For homeowners, property buyers, and anyone holding a shortlist

Most roofing materials are eliminated before price is discussed.

Steep-slope and residential low-slope · single-family and small multifamily

Slope, structure, climate, and repair access each rule candidates out. The matrix below runs every mainstream material past all four, and marks every cell where no defensible number exists rather than filling it in.

30-second answer

Which roofing material should I choose?

There is no best material, and this matrix is not a scoreboard. Slope, structural capacity, climate exposure, and whether anyone nearby can repair it eliminate most candidates before cost enters. Work those four gates in order. What usually survives is two or three materials — and only then does price decide between them.

Stop — immediate hazard

A roof that is already sagging, cracking, or letting daylight through is a structural question, not a material-choice question.

Every heavy material in the matrix below — tile, slate, and the heavier composites — adds permanent load to framing that was sized for something else. If there is already evidence that the existing structure is struggling, adding weight is the wrong conversation and choosing a covering is the wrong next step.

Nothing on this page requires you to get onto the roof or into the attic. Slope, condition, and geometry are all answerable from the ground, from an upstairs window, from photographs an installer takes for you, or from the drawings in a property file. OSHA records that “falls are the leading cause of death for workers engaged in residential construction,” and that trained workers must use conventional fall protection at six feet or more above lower levels. That is the reason to stay off, not a procedure to copy.

Keep people out from under the affected area and call a licensed structural engineer or your local building department. If a ceiling is bulging, a ridge line has visibly dropped, or you can hear cracking, leave the building and call 911.

Learning paths and saved lessons

At a glance

The four gates that eliminate, and the fifth that decidesSection link

Read the matrix column by column, not row by row. A material that fails an earlier gate cannot be rescued by a later one, and price is the last gate, not the first.

Gate 1 — Slope: Fixed by the building. Not negotiable without reframing.Below 2:12 the roof is held watertight by a continuous sealed surface rather than by overlap, and no shingle, tile, slate, or shake belongs there. The model-code minimum for every family is in the diagram and the matrix.
Gate 2 — Structure: Fixed by the framing. Answered by an engineer, not a table.Existing-building code sets a screening trigger rather than a weight limit. Section 3407.6.2 of the 2022 Oregon Structural Specialty Code — the edition Oregon superseded when the 2025 code took effect on 1 October 2025 — required a gravity-load-carrying element to be replaced or altered where an alteration raised its design gravity load by more than 5 percent, excepting a second layer of covering weighing 3 pounds per square foot or less. Quoted as a worked example, not as law anywhere today.
Gate 3 — Climate: Fit and misfit are different questions, and both matter.The matrix carries both. A material that performs well in one hazard often performs badly in another, and the misfit column is the one that catches a bad choice.
Gate 4 — Repair access: Who can fix it, in your market, in ten years?Service life is bounded by repairability. A century material with no local tradesperson is a shorter roof than a thirty-year material with twenty installers in town.
Gate 5 — Cost: Last, and not published here as a per-material numberThis site has no per-material installed-cost dataset it is willing to publish. Its own cost methodology page says that only the asphalt shingle rate behind its beta estimator is traceable to a source, that the other five material rates are unattributed beta values, and that they should not be quoted. So this page does not quote them.
What the matrix is not: Not a ranking, not a recommendation, not a code determinationEvery code figure in it is model text, read in the adopted code of a jurisdiction named in the source list with its edition and effective date. It is not the law where you live unless your jurisdiction has adopted that edition unamended. Confirm your adopted edition, its amendments, and its effective date with your authority having jurisdiction.

Tradeoffs

This page's advice — work the gates in order — and where that advice is wrongSection link

The position here is that slope, structure, climate, and repair access should settle the shortlist before anyone talks about price or appearance. That sequence is wrong often enough to be worth naming.

Best when

You are replacing a roof at end of life, on your own timetable, and genuinely have a choice of material.
The building is a candidate for something other than what is on it now — a re-cover, a material change, or a design change that alters slope.
You are comparing proposals that name different materials, and you need to know which differences are real constraints and which are preference.
You are buying a property and want to know what the existing roof commits you to, and what it forecloses.
The ownership horizon is long enough that replacement frequency, not first cost, is the number that decides.

Think twice if

The roof is leaking now. Triage comes first: an active leak is a diagnosis-and-stabilization problem, and material selection is a decision for after the building is dry.
A historic district, a design review board, or a homeowners' association controls appearance. Then the approved palette is the first gate, not the fourth, and the matrix is a way of checking what is left inside it.
An insurance claim is paying. The carrier's scope, deductible structure, and depreciation treatment can dominate the arithmetic, and an insurance scope is not the same document as a full replacement scope.
You expect to sell within a few years. Replacement-frequency logic assumes you own the roof through its life; if you do not, first cost and marketability carry more weight than service life.
Solar is planned. Array attachment, warranty interaction, and the cost of removing and resetting panels at the next re-roof couple the two decisions, and the covering choice should be made with the array in view.
The house predates 1990 and the existing covering, felts, or mastics may contain asbestos. Then the first question is testing and disturbance, not selection. The EPA recommends testing suspect materials where they are damaged or where planned work would disturb them, with sampling by a properly trained and accredited professional.

What changes the answer

The measured slope of every plane, not the slope of the biggest one. A porch, dormer, or addition at 1:12 needs a different system from the 8:12 main roof above it, and the two are priced and warranted separately.
Whether an engineer has actually looked at the framing, or whether 'the structure is fine' is a sentence someone said on the phone.
The dominant hazard where the building stands: hail frequency, design wind, wildfire exposure, snow load, freeze-thaw cycling, or salt. Each rewards a different column.
How many installers within a reasonable travel distance work in the material, and whether replacement units will still be produced in twenty years.
Whether the deck is coming off. A tear-off is the cheapest moment in a building's life to change material, because the labour to expose the deck has already been paid for.
The adopted code edition and its local amendments, which decide layer limits, ice-barrier extent, underlayment, and attachment — none of which are uniform across the United States.

The mechanism

Slope is a physics boundary, not a preferenceSection link

Everything else in the matrix is a tradeoff. This one is a wall.

Minimum slope for each roofing family, on one axis. Each bar starts at the model-code minimum and runs right, meaning that slope and steeper. The dashed line at 2:12 is where the physics changes. Every figure is model-code text as reproduced for the adopting jurisdiction named in the sources; it is not the law where you live until your jurisdiction adopts it.Original diagram, Understanding Roofing.

A steep-slope covering — shingle, shake, tile, slate, metal shingle — is a set of discrete pieces laid so each course laps the one below it. Nothing in that arrangement is sealed. It works because gravity carries a drop from lap to lap until it leaves at the eave, and it stops working when the water stops moving fast enough to clear the laps before it finds a sideways path. That is what a minimum slope is: the angle below which the geometry stops being self-draining.

A low-slope roof solves the same problem differently. Instead of overlap, it uses a continuous surface with welded, adhered, or torched seams — a membrane — that holds water out even when water sits still on it. The Department of Energy’s Building America Solution Center draws the boundary at the same place the trade does: low-slope roof assemblies are those with a “slope less than 2:12.”

Why the minimums do not line up neatly

The thresholds on the diagram are not a smooth ranking of quality. They are the outcome of how each system keeps water out:

0.25:12 for single-ply and standing seam. A fully welded membrane and a standing seam metal panel both have continuous, raised, or sealed joints, so the code minimum exists for drainage rather than for lap performance. The model text for built-up, modified bitumen, thermoset, and thermoplastic membranes all use the same words: a design slope of not less than one-fourth unit vertical in 12 units horizontal for drainage.
0.5:12 and 3:12 for the same metal panel. A lapped, non-soldered metal panel is permitted at one-half in twelve with applied lap sealant and only at three in twelve without it. The same sheet of metal has two different minimums depending on one line item in the scope. That is the clearest illustration on this page that a material is never the thing being specified — an assembly is.
2:12 for asphalt, 4:12 for slate. Thicker, more irregular, more permeable units need more gravity to clear their laps. Slate carries the highest minimum in the residential model code.
3:12 or 4:12 for wood shakes, depending on which model code you are reading. The residential model code sets shakes at 3:12 (IRC R905.8.2); the commercial model code sets them at 4:12 (IBC 1507.9.2). This page carries the residential figure because its subject is houses. It is the clearest reminder on the page that a minimum-slope number is not a fact about a material — it is a fact about a document, and which document governs is decided where the building stands.

The minimum is waived when you are replacing, not building

One qualification matters more on this page than anywhere else, because almost everyone reading a material matrix is re-roofing rather than building. NRCA’s technical director records that for replacement and re-cover of existing low-slope roofs, the model codes waive the 1/4:12 minimum and apply a performance requirement instead: positive drainage. The diagram above is therefore a new-construction ladder. On an existing low-slope roof that already drains, the question your jurisdiction will ask is whether it still drains — not whether it meets a slope your building never had.

The three gates behind slope

Slope is checkable from the ground. The three gates behind it are not, and each is a different kind of question.

Structure is arithmetic done by a person with a licence. There is no national weight limit for a roof covering, and the residential code does not set one; what exists instead is a trigger in existing-building code. Section 3407.6.2 of the 2022 Oregon Structural Specialty Code — the edition Oregon superseded when the 2025 code took effect on 1 October 2025 — required that “any existing gravity load-carrying structural element for which an alteration causes an increase in design gravity load of more than 5 percent shall be replaced or altered as needed to carry the design gravity loads required by this code for new structures,” with an exception for “buildings in which the increased dead load is due entirely to the addition of a second layer of roof covering weighing 3 pounds per square foot… or less over an existing single layer of roof covering.” Oregon has since adopted the 2025 Structural Specialty Code, built on the 2024 IBC and IEBC, and that edition — not the sentence quoted here — is what an Oregon plans examiner reads today. The provision is quoted as a worked example of the shape such a trigger takes, not as a rule in force anywhere.

Five percent of the load on one rafter is not a number you can look up. It is a calculation about your building, and the threshold itself has moved between model editions: the engineering firm Wiss, Janney, Elstner Associates, writing about the 2018 International Existing Building Code, records that the 2018 text counts an increase in design dead, live or snow load of more than 5 percent, where earlier editions counted only additional dead loads that increased forces in structural elements by more than 5 percent. Same number, different thing being measured. That is exactly the kind of difference an engineer is paid to notice and a table cannot carry.

Climate is two questions, not one. A material can be an excellent answer to the hazard you thought about and a poor answer to the one you did not. The matrix carries a fit column and a misfit column for exactly that reason, and the misfit column is the one that changes decisions.

Repair access is a market question that behaves like a technical one. The National Park Service, writing about historic slate, is blunt about the consequence of thin repair labour: repairs should be attempted only where a slater can be found, “because even the most experienced slater will likely damage additional slates while attempting repairs.” A material whose repair requires a trade your market does not have is a material that will be replaced rather than repaired, and that changes its effective service life more than any product property does.

The matrix

Every mainstream roofing material against the columns that decideSection link

Fifteen rows, nine columns. Every cell carrying a figure traces to a source named in the source list at the foot of this page; every cell with no defensible figure says so and names the variables instead. Rows are ordered by minimum slope, because that is the gate that eliminates first.

Roofing materials against minimum slope, structure, service life, cost basis, climate fit and misfit, repairability, and fire and impact listings. Slope figures are model-code text as adopted in a named jurisdiction; service-life figures are HUD capital-planning constants; cost cells deliberately carry no dollar figure.
Material	Full guide	Minimum slope, model code	Weight and structure gate	Service-life basis (HUD EUL)	Installed cost basis	Climate fit and misfit	Repairability	Fire and impact listings
Built-up roofing (BUR)	Built-up roofing	1/4:12 design slope for drainage; 1/8:12 for coal-tar BUR.	Multi-ply asphalt and felts plus surfacing is among the heavier low-slope options, and gravel surfacing adds more. Weight varies by ply count and surfacing; get it from the specified assembly, not from a category.	20 years with a gravel finish; 10 years without a mineral or gravel surfacing.	No figure published here. Driven by ply count, surfacing, insulation thickness, deck condition, and the disruption cost of hot work over an occupied building.	Fits: long service records on institutional buildings; redundancy from multiple plies. Misfits: hot application near occupied space; no reflectivity benefit unless surfaced for it.	Patchable ply by ply by a crew that still works in BUR. Finding that crew is the constraint in many residential markets.	Class A achievable as a tested assembly with the correct surfacing and deck. No impact classification applies; UL 2218 and FM 4473 are shingle tests.
Modified bitumen	Modified bitumen	1/4:12 design slope for drainage.	Lighter than BUR, heavier than most single-ply. Product-specific; take it from the membrane data sheet and the insulation schedule.	No separate HUD row. The nearest analogues are built-up roofing at 10 to 20 years and single-ply at 15.	No figure published here. Driven by application method (torch, cold-adhesive, self-adhered), number of plies, insulation, and access.	Fits: residential low-slope where a granulated cap surface and ply redundancy are wanted. Misfits: torch application adjacent to combustible construction; ponding.	Genuinely repairable with a compatible patch by a competent crew. Torch-applied work needs a trade with the right training and insurance.	Class A achievable as a tested assembly. No impact class.
EPDM single-ply	Low-slope roofing on houses	1/4:12 design slope for drainage (thermoset single-ply).	Among the lightest coverings available. Weight is dominated by the insulation and cover board below it rather than by the membrane.	15 years for an adhered rubber membrane.	No figure published here. Driven by attachment method (adhered, mechanically attached, ballasted), insulation thickness, and the number of penetrations.	Fits: the longest field record of the single plies; black EPDM absorbs solar heat, which helps clear snow and is forgiving in heating-dominated climates. Misfits: hot climates, where a dark membrane raises deck temperature; ponding; grease exposure.	Seams and punctures patch readily with compatible materials. Older seam technologies are the usual failure point, not the field.	Class A achievable as a tested assembly. No impact class.
TPO and PVC single-ply	TPO vs EPDM vs PVC	1/4:12 design slope for drainage (thermoplastic single-ply).	Among the lightest coverings available; the insulation and cover board dominate the load.	15 years for a thermoplastic membrane (TPO, vinyl). A rubberized or elastomeric white cool roof also carries 15.	No figure published here. Driven by membrane thickness, attachment, insulation, and penetration count.	Fits: hot climates, where a reflective white membrane lowers deck temperature; PVC where grease or chemical exposure is present. Misfits: cold, heating- dominated climates, where DOE notes reflective roofs can raise energy costs; ponding.	Heat-welded seams make patches strong and fast for a crew with a welder and a probe. The weld quality is the whole repair.	Class A achievable as a tested assembly. No impact class.
Standing seam metal	Standing seam metal	1/4:12 for standing-seam metal panel systems.	Usually the lightest durable steep-slope covering, which is why it is often the answer where the structure is marginal. The figure is product-specific: steel, aluminium, zinc and copper differ, and so do gauges.	50 years for metal. DOE separately states a metal roof can last up to 50 years or more.	No figure published here. DOE states only that metal roofing tends to cost more than asphalt shingles. Panel metal, gauge, coating, seam type, clip system, and the number of penetrations move it.	Fits: snow shedding, wildfire exposure, high wind with the right attachment, and long spans of simple geometry. Misfits: complex geometry with many penetrations; coastal salt without the right metal and fasteners — DOE calls for stainless fasteners or clips within 3,000 feet of a saltwater coastline.	Panels usually run eave to ridge, so replacing one means unclipping its neighbours. Coatings fade and a replacement panel will not match an aged one. Repainting is possible; matching is not.	Class A as an assembly, but DOE notes metal roofs usually need a noncombustible deck or a fire-resistant underlayment to qualify. Impact classes exist for some metal products; denting is cosmetic damage the impact test does not measure.
Lapped metal panel with sealed laps	Exposed-fastener metal	1/2:12 for lapped, non-soldered seam panels with applied lap sealant.	Light. Same product-specific caveat as standing seam.	Covered by the 50-year metal row, but the sealant in the lap is a maintenance item with a much shorter life than the panel.	No figure published here. Generally the least expensive metal option, and the difference is mostly labour.	Fits: outbuildings, porches and simple planes where the slope is too low for an unsealed lap. Misfits: anywhere the lap sealant will not be inspected and renewed.	Panels are removable, but exposed fasteners and lap sealant make the roof a maintenance schedule rather than an install-and-forget assembly.	Same assembly-level fire answer as other metal. No universal impact class.
Mineral-surfaced roll roofing	Low-slope roofing options	Not applied below 1:12.	Very light. Rarely a structural question.	No HUD row. Widely regarded as a short-life covering; no defensible figure found.	No figure published here. The cheapest covering per square in most markets, and the shortest-lived.	Fits: outbuildings, temporary work, and low-value structures. Misfits: anything you intend to keep; UV exposure and thermal movement age it fast.	Patchable, but the patch ages differently from the field and the whole covering reaches end of life at once.	Fire classification depends on the specific product and assembly. No impact class.
Asphalt shingles — architectural	Asphalt shingles	2:12 and steeper. From 2:12 up to 4:12, the model code calls for a double underlayment application.	Light enough that structure is rarely the gate on a house framed for shingles. A second layer over an existing single layer is the case the 3 pounds per square foot exception was written for — but adopted layer limits, not weight alone, decide whether a second layer is permitted.	20 years. This is the number most often confused with a warranty term.	No figure published here. The reference point most other materials are priced against. Product tier, geometry, tear-off layers, deck condition, and access move it.	Fits: almost every steep-slope climate, with the right wind class and underlayment; the widest installer base in the country. Misfits: sustained extreme heat and high UV, which accelerate ageing; severe hail belts, where even Class 4 product ages out of its class.	The easiest material on this page to repair, and the hardest to match. Granule colour shifts within a few years, so a repair is visible.	DOE states most asphalt shingle roofs carry a Class A fire rating without requiring a noncombustible deck. Wind classes under ASTM D3161 or D7158; impact classes 1 to 4 under UL 2218 where the product is rated.
Clay tile	Clay and concrete tile	2 1/2:12 and steeper; double underlayment from 2 1/2:12 to 4:12. Must be installed over wood structural panels or solid lumber sheathing.	Heavy. NPS states plainly that clay tiles are heavy and it is important that the roof structure be sound. Weight is profile-specific; a pan-and-cover roof is considerably heavier than an interlocking tile. No verified per-square figure is published here for clay.	60 years for clay or cementitious barrel tile.	No figure published here. Labour-intensive, with a thin installer base in most of the country outside the Southwest, Florida, and California.	Fits: hot-dry and hot-humid climates; long service records in the Southwest and Gulf. Misfits: freeze-thaw belts. NPS notes porous tiles are particularly susceptible to breaking and spalling during freeze-thaw and can let moisture reach battens and structure.	Individual tiles are replaceable, and walking the roof to reach them breaks more. Discontinued profiles are a real matching risk. The underlayment, not the tile, is doing the waterproofing, and it fails first.	Deemed Class A on noncombustible decks under the model code exceptions; otherwise the assembly must be tested. No standard impact class; tiles crack rather than dent.
Concrete tile	Clay and concrete tile	Same as clay: 2 1/2:12 and steeper, over solid sheathing.	The one weight figure this page publishes. One manufacturer's currently published technical page lists standard-weight profiles at roughly 900 to 970 pounds per 100 square feet, and lightweight profiles at roughly 550 to 720 — a difference large enough to change whether a building qualifies. Use the data sheet for the specific product proposed.	Covered by the 60-year clay or cementitious barrel tile row.	No figure published here. Generally less than clay for the tile itself; the labour and structural implications are similar.	Fits: hot-dry climates; wind performance is good with the right attachment. Misfits: freeze-thaw, and any building where the structural evaluation has not been done.	As clay: individually replaceable, breakable underfoot, colour shifts with weathering, profiles get discontinued.	Deemed Class A on noncombustible decks under the model code exceptions. No standard impact class.
Metal shingles and stone-coated steel	Metal shingles and stone-coated steel	Not installed on roof slopes below 3:12.	Light, like other metal, and the usual reason it is specified over tile or slate on framing that cannot carry them. Stone-coated products are heavier than bare panel; the figure is product-specific.	Covered by the 50-year HUD metal row, which does not distinguish panel from shingle.	No figure published here. Between panel metal and tile in most markets; the labour is closer to shingle work than to standing seam.	Fits: buildings that want a tile or shake appearance without the weight; wind performance is good with the right attachment. Misfits: coastal salt without the right metal and fasteners; markets where nobody stocks the profile for a repair.	Individually replaceable in principle. The matching problem is severe: stone-coated granules weather, and a discontinued profile ends repairability.	Deemed Class A on noncombustible decks under the model code exceptions; on a combustible deck the assembly generally needs a fire-resistant underlayment. Impact classes exist for some products; denting is not what the impact test measures.
Wood shingles	Wood shingles and shakes	3:12 and steeper.	Light. NPS notes wood shingles were historically preferred over tile in part because they were much lighter and required less heavy and less expensive roof framing.	25 years for wood shingle and cedar shakes or shingles.	No figure published here. Material cost is grade-dependent and the maintenance obligation is a recurring cost, not a one-off.	Fits: regions with an established cedar tradition and a trade to maintain it. Misfits: wildfire-exposed areas, where many jurisdictions restrict or prohibit them; damp shaded sites where moss and rot shorten life sharply.	Individually replaceable by a trade that still exists in some regions and not others. New shingles will not match weathered ones for years.	Fire classification depends entirely on treatment and the tested assembly, and some jurisdictions restrict wood roofing outright. No standard impact class.
Wood shakes	Wood shingles and shakes	3:12 and steeper under the residential model code (IRC R905.8.2) — the same minimum as wood shingles. The commercial model code sets 4:12 for shakes (IBC 1507.9.2). This is the one row where the two model codes disagree, so the adopted edition matters more than the number.	Light, like wood shingles.	25 years, in the same HUD row as wood shingles.	No figure published here. Higher than wood shingles for material and labour, with the same recurring maintenance obligation.	Fits: regional traditions with a maintaining trade. Misfits: wildfire exposure, and damp shaded sites. Restrictions on wood roofing are common and jurisdiction-specific.	Individually replaceable; matching is poor for years. The trade is regional.	As wood shingles: classification depends on treatment and the tested assembly, and some jurisdictions restrict wood roofing.
Natural slate	Natural slate	4:12 and steeper — the highest minimum on this page. Fastened to wood structural panels or solid lumber sheathing.	Heavy, and a structural evaluation is the gate. NPS instructs that roof timbers be checked for deflection, cracking and twisting, and that a structural engineer experienced with older buildings be consulted where such conditions are found. No verified per-square weight figure is published here.	75 years in the HUD table. NPS reports 60 to 125 years or longer, with about 125 for Vermont and New York slates, 175 or more for Buckingham Virginia, and in excess of 60 for Pennsylvania Soft-Vein — and cautions these are a general guide only.	No figure published here. Labour-intensive, with the thinnest installer base of any material on this page in most markets.	Fits: almost any climate, given the right slate and the right fasteners; the longest documented record of any covering here. Misfits: buildings whose framing has not been evaluated, and markets with no slater — NPS warns even an experienced slater will damage additional slates while making repairs.	Repairable indefinitely in principle, and salvaged slates can be reused. In practice repairability is set by whether a slater is reachable. Fasteners and flashings fail long before the stone does.	Deemed Class A on noncombustible decks, and also over underlayment on combustible decks, under the model code exceptions. No standard impact class.
Composite (polymer) slate and shake	Composite roofing	Set by the product, not by a code family — which is why this row sits last rather than in slope order. The published instructions for the specific product govern, and they may set a higher minimum than the family the product imitates.	Markedly lighter than the stone or tile it imitates, which is the principal engineering reason to choose it. Weight is product-specific and no verified figure is published here.	No HUD row exists for this category. It is genuinely young, and no source found offers a defensible multi-decade figure. NPS observes that over the long term several artificial slate roofs will have to be installed during the life span of one natural slate roof.	No figure published here. NPS notes that because labour, flashings and removal costs are equal, the total initial cost of an artificial slate roof is only marginally less than a natural slate roof.	Fits: buildings that want the appearance and cannot carry the weight, and markets with no slater. Misfits: restoration work — NPS does not recommend artificial mineral-fiber slate for it, and notes artificial slates may tend to fade.	Individually replaceable and lighter to handle. The real risks are colour drift, which NPS names, and profile discontinuation in a category where products change often.	Many products carry Class A fire and Class 4 impact listings, always as part of a specified assembly. Read the product's own listing rather than the category.

Read this table one item at a time

Built-up roofing (BUR)

Full guide: Built-up roofing
Minimum slope, model code: 1/4:12 design slope for drainage; 1/8:12 for coal-tar BUR.
Weight and structure gate: Multi-ply asphalt and felts plus surfacing is among the heavier low-slope options, and gravel surfacing adds more. Weight varies by ply count and surfacing; get it from the specified assembly, not from a category.
Service-life basis (HUD EUL): 20 years with a gravel finish; 10 years without a mineral or gravel surfacing.
Installed cost basis: No figure published here. Driven by ply count, surfacing, insulation thickness, deck condition, and the disruption cost of hot work over an occupied building.
Climate fit and misfit: Fits: long service records on institutional buildings; redundancy from multiple plies. Misfits: hot application near occupied space; no reflectivity benefit unless surfaced for it.
Repairability: Patchable ply by ply by a crew that still works in BUR. Finding that crew is the constraint in many residential markets.
Fire and impact listings: Class A achievable as a tested assembly with the correct surfacing and deck. No impact classification applies; UL 2218 and FM 4473 are shingle tests.

Modified bitumen

Full guide: Modified bitumen
Minimum slope, model code: 1/4:12 design slope for drainage.
Weight and structure gate: Lighter than BUR, heavier than most single-ply. Product-specific; take it from the membrane data sheet and the insulation schedule.
Service-life basis (HUD EUL): No separate HUD row. The nearest analogues are built-up roofing at 10 to 20 years and single-ply at 15.
Installed cost basis: No figure published here. Driven by application method (torch, cold-adhesive, self-adhered), number of plies, insulation, and access.
Climate fit and misfit: Fits: residential low-slope where a granulated cap surface and ply redundancy are wanted. Misfits: torch application adjacent to combustible construction; ponding.
Repairability: Genuinely repairable with a compatible patch by a competent crew. Torch-applied work needs a trade with the right training and insurance.
Fire and impact listings: Class A achievable as a tested assembly. No impact class.

EPDM single-ply

Full guide: Low-slope roofing on houses
Minimum slope, model code: 1/4:12 design slope for drainage (thermoset single-ply).
Weight and structure gate: Among the lightest coverings available. Weight is dominated by the insulation and cover board below it rather than by the membrane.
Service-life basis (HUD EUL): 15 years for an adhered rubber membrane.
Installed cost basis: No figure published here. Driven by attachment method (adhered, mechanically attached, ballasted), insulation thickness, and the number of penetrations.
Climate fit and misfit: Fits: the longest field record of the single plies; black EPDM absorbs solar heat, which helps clear snow and is forgiving in heating-dominated climates. Misfits: hot climates, where a dark membrane raises deck temperature; ponding; grease exposure.
Repairability: Seams and punctures patch readily with compatible materials. Older seam technologies are the usual failure point, not the field.
Fire and impact listings: Class A achievable as a tested assembly. No impact class.

TPO and PVC single-ply

Full guide: TPO vs EPDM vs PVC
Minimum slope, model code: 1/4:12 design slope for drainage (thermoplastic single-ply).
Weight and structure gate: Among the lightest coverings available; the insulation and cover board dominate the load.
Service-life basis (HUD EUL): 15 years for a thermoplastic membrane (TPO, vinyl). A rubberized or elastomeric white cool roof also carries 15.
Installed cost basis: No figure published here. Driven by membrane thickness, attachment, insulation, and penetration count.
Climate fit and misfit: Fits: hot climates, where a reflective white membrane lowers deck temperature; PVC where grease or chemical exposure is present. Misfits: cold, heating- dominated climates, where DOE notes reflective roofs can raise energy costs; ponding.
Repairability: Heat-welded seams make patches strong and fast for a crew with a welder and a probe. The weld quality is the whole repair.
Fire and impact listings: Class A achievable as a tested assembly. No impact class.

Standing seam metal

Full guide: Standing seam metal
Minimum slope, model code: 1/4:12 for standing-seam metal panel systems.
Weight and structure gate: Usually the lightest durable steep-slope covering, which is why it is often the answer where the structure is marginal. The figure is product-specific: steel, aluminium, zinc and copper differ, and so do gauges.
Service-life basis (HUD EUL): 50 years for metal. DOE separately states a metal roof can last up to 50 years or more.
Installed cost basis: No figure published here. DOE states only that metal roofing tends to cost more than asphalt shingles. Panel metal, gauge, coating, seam type, clip system, and the number of penetrations move it.
Climate fit and misfit: Fits: snow shedding, wildfire exposure, high wind with the right attachment, and long spans of simple geometry. Misfits: complex geometry with many penetrations; coastal salt without the right metal and fasteners — DOE calls for stainless fasteners or clips within 3,000 feet of a saltwater coastline.
Repairability: Panels usually run eave to ridge, so replacing one means unclipping its neighbours. Coatings fade and a replacement panel will not match an aged one. Repainting is possible; matching is not.
Fire and impact listings: Class A as an assembly, but DOE notes metal roofs usually need a noncombustible deck or a fire-resistant underlayment to qualify. Impact classes exist for some metal products; denting is cosmetic damage the impact test does not measure.

Lapped metal panel with sealed laps

Full guide: Exposed-fastener metal
Minimum slope, model code: 1/2:12 for lapped, non-soldered seam panels with applied lap sealant.
Weight and structure gate: Light. Same product-specific caveat as standing seam.
Service-life basis (HUD EUL): Covered by the 50-year metal row, but the sealant in the lap is a maintenance item with a much shorter life than the panel.
Installed cost basis: No figure published here. Generally the least expensive metal option, and the difference is mostly labour.
Climate fit and misfit: Fits: outbuildings, porches and simple planes where the slope is too low for an unsealed lap. Misfits: anywhere the lap sealant will not be inspected and renewed.
Repairability: Panels are removable, but exposed fasteners and lap sealant make the roof a maintenance schedule rather than an install-and-forget assembly.
Fire and impact listings: Same assembly-level fire answer as other metal. No universal impact class.

Mineral-surfaced roll roofing

Full guide: Low-slope roofing options
Minimum slope, model code: Not applied below 1:12.
Weight and structure gate: Very light. Rarely a structural question.
Service-life basis (HUD EUL): No HUD row. Widely regarded as a short-life covering; no defensible figure found.
Installed cost basis: No figure published here. The cheapest covering per square in most markets, and the shortest-lived.
Climate fit and misfit: Fits: outbuildings, temporary work, and low-value structures. Misfits: anything you intend to keep; UV exposure and thermal movement age it fast.
Repairability: Patchable, but the patch ages differently from the field and the whole covering reaches end of life at once.
Fire and impact listings: Fire classification depends on the specific product and assembly. No impact class.

Asphalt shingles — architectural

Full guide: Asphalt shingles
Minimum slope, model code: 2:12 and steeper. From 2:12 up to 4:12, the model code calls for a double underlayment application.
Weight and structure gate: Light enough that structure is rarely the gate on a house framed for shingles. A second layer over an existing single layer is the case the 3 pounds per square foot exception was written for — but adopted layer limits, not weight alone, decide whether a second layer is permitted.
Service-life basis (HUD EUL): 20 years. This is the number most often confused with a warranty term.
Installed cost basis: No figure published here. The reference point most other materials are priced against. Product tier, geometry, tear-off layers, deck condition, and access move it.
Climate fit and misfit: Fits: almost every steep-slope climate, with the right wind class and underlayment; the widest installer base in the country. Misfits: sustained extreme heat and high UV, which accelerate ageing; severe hail belts, where even Class 4 product ages out of its class.
Repairability: The easiest material on this page to repair, and the hardest to match. Granule colour shifts within a few years, so a repair is visible.
Fire and impact listings: DOE states most asphalt shingle roofs carry a Class A fire rating without requiring a noncombustible deck. Wind classes under ASTM D3161 or D7158; impact classes 1 to 4 under UL 2218 where the product is rated.

Clay tile

Full guide: Clay and concrete tile
Minimum slope, model code: 2 1/2:12 and steeper; double underlayment from 2 1/2:12 to 4:12. Must be installed over wood structural panels or solid lumber sheathing.
Weight and structure gate: Heavy. NPS states plainly that clay tiles are heavy and it is important that the roof structure be sound. Weight is profile-specific; a pan-and-cover roof is considerably heavier than an interlocking tile. No verified per-square figure is published here for clay.
Service-life basis (HUD EUL): 60 years for clay or cementitious barrel tile.
Installed cost basis: No figure published here. Labour-intensive, with a thin installer base in most of the country outside the Southwest, Florida, and California.
Climate fit and misfit: Fits: hot-dry and hot-humid climates; long service records in the Southwest and Gulf. Misfits: freeze-thaw belts. NPS notes porous tiles are particularly susceptible to breaking and spalling during freeze-thaw and can let moisture reach battens and structure.
Repairability: Individual tiles are replaceable, and walking the roof to reach them breaks more. Discontinued profiles are a real matching risk. The underlayment, not the tile, is doing the waterproofing, and it fails first.
Fire and impact listings: Deemed Class A on noncombustible decks under the model code exceptions; otherwise the assembly must be tested. No standard impact class; tiles crack rather than dent.

Concrete tile

Full guide: Clay and concrete tile
Minimum slope, model code: Same as clay: 2 1/2:12 and steeper, over solid sheathing.
Weight and structure gate: The one weight figure this page publishes. One manufacturer's currently published technical page lists standard-weight profiles at roughly 900 to 970 pounds per 100 square feet, and lightweight profiles at roughly 550 to 720 — a difference large enough to change whether a building qualifies. Use the data sheet for the specific product proposed.
Service-life basis (HUD EUL): Covered by the 60-year clay or cementitious barrel tile row.
Installed cost basis: No figure published here. Generally less than clay for the tile itself; the labour and structural implications are similar.
Climate fit and misfit: Fits: hot-dry climates; wind performance is good with the right attachment. Misfits: freeze-thaw, and any building where the structural evaluation has not been done.
Repairability: As clay: individually replaceable, breakable underfoot, colour shifts with weathering, profiles get discontinued.
Fire and impact listings: Deemed Class A on noncombustible decks under the model code exceptions. No standard impact class.

Metal shingles and stone-coated steel

Full guide: Metal shingles and stone-coated steel
Minimum slope, model code: Not installed on roof slopes below 3:12.
Weight and structure gate: Light, like other metal, and the usual reason it is specified over tile or slate on framing that cannot carry them. Stone-coated products are heavier than bare panel; the figure is product-specific.
Service-life basis (HUD EUL): Covered by the 50-year HUD metal row, which does not distinguish panel from shingle.
Installed cost basis: No figure published here. Between panel metal and tile in most markets; the labour is closer to shingle work than to standing seam.
Climate fit and misfit: Fits: buildings that want a tile or shake appearance without the weight; wind performance is good with the right attachment. Misfits: coastal salt without the right metal and fasteners; markets where nobody stocks the profile for a repair.
Repairability: Individually replaceable in principle. The matching problem is severe: stone-coated granules weather, and a discontinued profile ends repairability.
Fire and impact listings: Deemed Class A on noncombustible decks under the model code exceptions; on a combustible deck the assembly generally needs a fire-resistant underlayment. Impact classes exist for some products; denting is not what the impact test measures.

Wood shingles

Full guide: Wood shingles and shakes
Minimum slope, model code: 3:12 and steeper.
Weight and structure gate: Light. NPS notes wood shingles were historically preferred over tile in part because they were much lighter and required less heavy and less expensive roof framing.
Service-life basis (HUD EUL): 25 years for wood shingle and cedar shakes or shingles.
Installed cost basis: No figure published here. Material cost is grade-dependent and the maintenance obligation is a recurring cost, not a one-off.
Climate fit and misfit: Fits: regions with an established cedar tradition and a trade to maintain it. Misfits: wildfire-exposed areas, where many jurisdictions restrict or prohibit them; damp shaded sites where moss and rot shorten life sharply.
Repairability: Individually replaceable by a trade that still exists in some regions and not others. New shingles will not match weathered ones for years.
Fire and impact listings: Fire classification depends entirely on treatment and the tested assembly, and some jurisdictions restrict wood roofing outright. No standard impact class.

Wood shakes

Full guide: Wood shingles and shakes
Minimum slope, model code: 3:12 and steeper under the residential model code (IRC R905.8.2) — the same minimum as wood shingles. The commercial model code sets 4:12 for shakes (IBC 1507.9.2). This is the one row where the two model codes disagree, so the adopted edition matters more than the number.
Weight and structure gate: Light, like wood shingles.
Service-life basis (HUD EUL): 25 years, in the same HUD row as wood shingles.
Installed cost basis: No figure published here. Higher than wood shingles for material and labour, with the same recurring maintenance obligation.
Climate fit and misfit: Fits: regional traditions with a maintaining trade. Misfits: wildfire exposure, and damp shaded sites. Restrictions on wood roofing are common and jurisdiction-specific.
Repairability: Individually replaceable; matching is poor for years. The trade is regional.
Fire and impact listings: As wood shingles: classification depends on treatment and the tested assembly, and some jurisdictions restrict wood roofing.

Natural slate

Full guide: Natural slate
Minimum slope, model code: 4:12 and steeper — the highest minimum on this page. Fastened to wood structural panels or solid lumber sheathing.
Weight and structure gate: Heavy, and a structural evaluation is the gate. NPS instructs that roof timbers be checked for deflection, cracking and twisting, and that a structural engineer experienced with older buildings be consulted where such conditions are found. No verified per-square weight figure is published here.
Service-life basis (HUD EUL): 75 years in the HUD table. NPS reports 60 to 125 years or longer, with about 125 for Vermont and New York slates, 175 or more for Buckingham Virginia, and in excess of 60 for Pennsylvania Soft-Vein — and cautions these are a general guide only.
Installed cost basis: No figure published here. Labour-intensive, with the thinnest installer base of any material on this page in most markets.
Climate fit and misfit: Fits: almost any climate, given the right slate and the right fasteners; the longest documented record of any covering here. Misfits: buildings whose framing has not been evaluated, and markets with no slater — NPS warns even an experienced slater will damage additional slates while making repairs.
Repairability: Repairable indefinitely in principle, and salvaged slates can be reused. In practice repairability is set by whether a slater is reachable. Fasteners and flashings fail long before the stone does.
Fire and impact listings: Deemed Class A on noncombustible decks, and also over underlayment on combustible decks, under the model code exceptions. No standard impact class.

Composite (polymer) slate and shake

Full guide: Composite roofing
Minimum slope, model code: Set by the product, not by a code family — which is why this row sits last rather than in slope order. The published instructions for the specific product govern, and they may set a higher minimum than the family the product imitates.
Weight and structure gate: Markedly lighter than the stone or tile it imitates, which is the principal engineering reason to choose it. Weight is product-specific and no verified figure is published here.
Service-life basis (HUD EUL): No HUD row exists for this category. It is genuinely young, and no source found offers a defensible multi-decade figure. NPS observes that over the long term several artificial slate roofs will have to be installed during the life span of one natural slate roof.
Installed cost basis: No figure published here. NPS notes that because labour, flashings and removal costs are equal, the total initial cost of an artificial slate roof is only marginally less than a natural slate roof.
Climate fit and misfit: Fits: buildings that want the appearance and cannot carry the weight, and markets with no slater. Misfits: restoration work — NPS does not recommend artificial mineral-fiber slate for it, and notes artificial slates may tend to fade.
Repairability: Individually replaceable and lighter to handle. The real risks are colour drift, which NPS names, and profile discontinuation in a category where products change often.
Fire and impact listings: Many products carry Class A fire and Class 4 impact listings, always as part of a specified assembly. Read the product's own listing rather than the category.

Slope figures are model-code text, read in the adopted code of a jurisdiction named in the source list with its edition and effective date; confirm the edition adopted where the building stands. They are new-construction minimums: for replacement and re-cover of an existing low-slope roof, the model codes waive the 1/4:12 minimum and apply a positive-drainage performance requirement instead. Service-life figures are HUD Estimated Useful Life constants for multifamily capital needs assessment — planning numbers chosen so a reserve is funded in time, not predictions for one house. “No figure published here” means exactly that: this site holds no cost dataset it is willing to publish, and an honest blank is preferable to an unsourced range.

Instead of a sort control

The three orderings that actually change a decisionSection link

This table has no sort buttons. A sortable table needs JavaScript to be useful and needs careful work to stay usable without it, and the three orderings a reader genuinely wants are short enough to publish outright. The matrix itself is ordered by minimum slope, because that is the gate that eliminates first.

By minimum slope, shallowest first

Coal-tar built-up (1/8:12) → other built-up, modified bitumen, EPDM, TPO and PVC, standing seam metal (1/4:12) → lapped metal with sealed laps (1/2:12) → mineral-surfaced roll roofing (1:12) → asphalt shingles (2:12) → clay and concrete tile (2 1/2:12) → metal shingles, lapped metal without sealant, wood shingles, wood shakes (3:12) → natural slate (4:12). Wood shakes sit at 3:12 because that is the residential model code; the commercial model code puts them at 4:12.

By HUD estimated useful life, longest first

Slate shingle (75) → clay or cementitious barrel tile (60) → metal (50) → wood shingle and cedar shakes (25) → asphalt shingle and built-up with gravel finish (20) → EPDM, thermoplastic membrane, and elastomeric cool roof (15) → built-up without mineral or gravel finish (10).

By how much the structure has to be checked

Lightest and least likely to trigger a structural question: single-ply membranes, standing seam and other metal, wood shingles and shakes, asphalt shingles. Middle: modified bitumen and built-up, where insulation and surfacing dominate. Heaviest and always a structural question on a material change: clay tile, concrete tile, and natural slate. Composite slate and shake sits deliberately in the light group, which is most of its argument.

Those three lists are the whole content of a sort control, published as text so they work with no script, no pointer, and no assistive-technology workaround. The matrix above scrolls horizontally inside its own container and is reachable from the keyboard as a labelled region.

Worked example

How many roofs you buy in sixty yearsSection link

One arithmetic step on one federal dataset. It is not a cost comparison, and it deliberately cannot become one — but it is the number that reframes what 'expensive' means.

The U.S. Department of Housing and Urban Development publishes an Estimated Useful Life table for the capital needs assessments that multifamily properties must produce. It assigns one figure to each roof covering family. HUD describes it as “the recommended average useful life of the categories of assets that should be considered in a Capital Needs Assessment,” and states that the standard figure is fixed but that an assessor may enter a different assessed remaining useful life with a justification.

The arithmetic below is ours, not HUD’s. Take a sixty-year hold — roughly a single long ownership, or one full capital planning cycle — and count how many coverings you buy across it, including the one you buy today. Fractions round up, because you cannot buy nine-tenths of a roof.

Coverings purchased across a sixty-year hold, computed from HUD Estimated Useful Life figures. The service-life figures are HUD's; the count and the interpretation are this page's.
HUD component	Estimated useful life	Coverings bought in 60 years	What that means in practice
Asphalt shingle	20 years	3	Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Metal	50 years	2	Bought 2 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Slate shingle	75 years	1	Bought once. The next covering decision belongs to someone else.
Clay/cementitious barrel tile	60 years	1	Bought once. The next covering decision belongs to someone else.
Wood shingle, cedar shakes/shingles	25 years	3	Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Low slope — built-up roof, with gravel finish	20 years	3	Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Low slope — built-up roof, no mineral or gravel finish	10 years	6	Bought 6 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Low slope — adhered rubber membrane (EPDM)	15 years	4	Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Low slope — thermoplastic membrane (TPO, vinyl)	15 years	4	Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.
Low slope — rubberized/elastomeric white/cool roof	15 years	4	Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Read this table one item at a time

Asphalt shingle

Estimated useful life: 20 years
Coverings bought in 60 years: 3
What that means in practice: Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Metal

Estimated useful life: 50 years
Coverings bought in 60 years: 2
What that means in practice: Bought 2 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Slate shingle

Estimated useful life: 75 years
Coverings bought in 60 years: 1
What that means in practice: Bought once. The next covering decision belongs to someone else.

Clay/cementitious barrel tile

Estimated useful life: 60 years
Coverings bought in 60 years: 1
What that means in practice: Bought once. The next covering decision belongs to someone else.

Wood shingle, cedar shakes/shingles

Estimated useful life: 25 years
Coverings bought in 60 years: 3
What that means in practice: Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Low slope — built-up roof, with gravel finish

Estimated useful life: 20 years
Coverings bought in 60 years: 3
What that means in practice: Bought 3 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Low slope — built-up roof, no mineral or gravel finish

Estimated useful life: 10 years
Coverings bought in 60 years: 6
What that means in practice: Bought 6 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Low slope — adhered rubber membrane (EPDM)

Estimated useful life: 15 years
Coverings bought in 60 years: 4
What that means in practice: Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Low slope — thermoplastic membrane (TPO, vinyl)

Estimated useful life: 15 years
Coverings bought in 60 years: 4
What that means in practice: Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Low slope — rubberized/elastomeric white/cool roof

Estimated useful life: 15 years
Coverings bought in 60 years: 4
What that means in practice: Bought 4 times. Every replacement is a full tear-off, disposal, deck inspection, and disruption event — not just the price of the covering.

Service-life figures: HUD CNA e-Tool Estimated Useful Life table. Counts and commentary: original to this page. Both HUD columns (family and elderly properties) carry the same figure for every roofing row.

What this example does not show

It is not a cost comparison. Whether three asphalt roofs cost more or less than one slate roof depends on prices this page does not publish, on how those prices move over sixty years, and on the time value of money. The count tells you how many purchase events there are, not what each one costs.
It assumes you never sell. Most owners do. If the hold is fifteen years, every material in the table is bought once and the argument evaporates.
The federal figures disagree with the other federal source on this page. HUD assigns slate 75 years. The National Park Service reports 60 to 125 years or longer for slate roofs, about 125 for Vermont and New York slates, and 175 or more for Buckingham Virginia. Neither is wrong. HUD is setting a reserve contribution so that money exists when it is needed; NPS is describing the observed durability of quarried stone. A number is only meaningful with the question it was built to answer attached to it.
It says nothing about failure risk. A roof that reaches its estimated useful life without incident and a roof that leaks in year seven both appear in this table as the same number. Installation quality, ventilation, and flashing detail decide which one you get, and none of them is a material property. Flashing in particular causes more leaks than field material does — see roof flashing.
There is no row for composite. HUD has none, and this page does not invent one. That absence is itself information about how young the category is.

Using the matrix

A shortlist of two or three, and the question that separates themSection link

The matrix is a filter, not an answer. Run it in order and it usually leaves two or three candidates, which is the point at which a real conversation with an installer becomes possible.

Measure the slope of every plane. Cross out every row whose minimum is above the shallowest plane you need to cover. If the roof has both a 9:12 main plane and a 1:12 porch, you are choosing two systems and a transition detail, not one material.
Decide whether the structure is a live question. If the shortlist still contains tile or slate, it is, and an engineer answers it before anything else happens. If it does not, weight probably drops out of the decision entirely.
Read the misfit column, not the fit column. The fit column tells you what a material is good at. The misfit column is where a decision goes wrong. Match it against the hazard that actually governs at the address — hail, wind, wildfire, snow, freeze-thaw, salt, or sustained heat.
Ask who repairs it here. Not who installs it — who repairs it, in ten years, for a two-tile job. If the honest answer is nobody, the service-life column has just become fiction for that row.
Only now, price the survivors. Same unit, same scope, same exclusions, in writing. A price comparison between materials is only meaningful once the shortlist is already correct.

If the roof is currently leaking, none of this is the first step. Stabilize first: see roof leaks for how water actually travels, emergency leaks and tarping for the first hours, and repair or replace for whether a material decision is even due yet.

Cost and lifecycle

What this page will not tell you about cost, and whySection link

The cost column in the matrix names variables instead of dollars. That is a deliberate refusal, and the basis below describes the figure this site would have to be able to defend before it published one.

Units: U.S. dollars per roofing square (100 sq ft) of roof surface, installed — the only unit in which two proposals for different materials can be compared
Scope included: Tear-off of the existing covering, deck inspection, underlayment, the covering itself, flashings, edge metal, ventilation components, disposal, and permit
Not included: Deck replacement, structural work, chimney and skylight work, gutters, insulation, access and staging premiums, and anything discovered after the deck is exposed
Geography: United States, national planning range — a national figure cannot describe any local market and is only useful for arguing with a proposal
Data as of: August 2026
Confidence: None published. This site holds no per-material installed-cost dataset that meets its own evidence standard. Its cost methodology page states that only the asphalt shingle rate behind its beta estimator is traceable to a recorded source, that the other five material rates are unattributed beta values, and that they are not evidence and should not be quoted.
Method: Read the published cost methodology, including what it admits it cannot support

Almost every roofing comparison table on the internet carries a dollar range for each material. Very few carry a source for it, and the ranges disagree with each other by factors of two or three. A number with no scope, no geography, no date, and no method is not evidence about your roof; it is a number that makes a table look finished.

There are two sourced, directional cost statements this page is willing to repeat, and both are relative rather than numeric. The Department of Energy’s Building America Solution Center says plainly that “metal roofing tends to cost more than asphalt shingles.” The National Park Service, comparing natural slate with its synthetic substitutes, observes that because labour, flashings, and removal costs are equal, “the total initial cost of an artificial slate roof is only marginally less than a natural slate roof,” and that “over the long term, natural slate tends to be a better investment because several artificial slate roofs will have to be installed during the life span of one natural slate roof.”

What actually moves a per-material price

Labour intensity per square, not material price. Slate, tile, and standing seam are slower to install than asphalt shingles by a wide margin, and in most markets labour is the larger half of a roofing bill.
How thin the local trade is. A material three companies in a metro will quote is priced differently from one thirty companies will quote. This is a local labour-market fact, not a property of the product.
What the material forces you to do to the building. Structural reinforcement, deck replacement, batten systems, and heavier flashing metals are triggered by the choice and are often quoted as separate lines — or omitted, which is worse.
Geometry. Valleys, dormers, hips, penetrations, and roof-to-wall runs multiply detail labour, and they multiply it more for slow materials than for fast ones. Two roofs of the same area in the same material can differ by half again on geometry alone.
Access and staging. Steepness, height, ground conditions, and where a truck can park change the price of the identical roof.

Until this site publishes a dataset it can stand behind, the useful move is not to look up a range but to make three proposals comparable: same unit, same scope, same exclusions. That is a different skill, and it is the one that actually saves money.

No figure on this page is a quote, and this page deliberately publishes none. The only price that binds anyone is the one in a signed scope of work for this building.

Considerations

What changes this on a real buildingSection link

Structural weight

Weight is the one column in the matrix that can make a material simply unavailable. The residential code does not publish a weight limit, and no table can, because the answer depends on the framing member, its span, its spacing, its species and grade, its condition, and the snow and wind loads already acting on it.

What exists instead is a trigger, and the trigger has been written more than one way. The engineering firm Wiss, Janney, Elstner Associates, writing for practitioners about the 2018 International Existing Building Code, quotes Section 706.2 as catching “any existing gravity load-carrying structural element for which an alteration causes an increase in design dead, live or snow load, including snow drift effects, of more than 5 percent,” and notes that editions before 2018 instead caught additional dead loads that increased forces in structural elements by more than 5 percent. WJE also states plainly that its citations are to the model 2018 IEBC and that local adopted provisions may differ. Both versions carry the same exception: a second layer of roof covering weighing 3 pounds per square foot or less over an existing single layer.

The National Park Service puts the practical version of this in one sentence about clay tile: “Clay tiles are heavy and it is important that the roof structure be sound.” Its slate brief goes further, telling readers to check roof timbers for “deflection, cracking, and twisting,” and that where such conditions are found “a structural engineer experienced in working with older buildings should be consulted.”

A change to a heavier covering is a structural question, answered by a licensed engineer for this building. No matrix, calculator, or contractor's assurance substitutes for that. The 5 percent figure is a screening trigger drawn from model existing-building text and one state's superseded adoption of it — not a national permission, not a current requirement anywhere on this page's authority, and not a substitute for a calculation. Confirm the adopted edition and its effective date with your authority having jurisdiction.

Slope and drainage

Measure every plane, not the roof. Houses with a low-slope porch, a shed dormer, or a flat-roofed addition routinely need two systems, and the junction between them is a detail rather than a seam. The pitch of each plane is checkable from the ground or from a photograph — see roof pitch for how, and roof measurement for the ground-based and document-based methods that avoid the roof entirely.

Minimum slopes in the matrix are model-code text as adopted somewhere. A specific product's published instructions may set a higher minimum than the code does, and where they do, the instructions govern the warranty.

Fire

Fire classification is a property of an assembly, and the model code says so in its own words. Section R902.1 of the 2024 International Residential Code — adopted by the Georgia Board of Community Affairs as the 2024 IRC with Georgia Amendments, effective 1 January 2026 — requires that “Class A, B or C roof assemblies shall be installed in jurisdictions designated by law as requiring their use or where the edge of the roof deck is less than 3 feet from a lot line,” and that “where Class A, B or C roof assemblies are required, they shall be tested in accordance with ASTM E108 or UL 790.”

The exceptions in that same section prove the point rather than weakening it. Metal sheets and shingles, clay and concrete tile, and slate are deemed Class A without testing on noncombustible decks; slate is also deemed Class A over underlayment on combustible decks; copper needs a minimum weight per square foot. The deck is part of the answer in every case. The Building America Solution Center states the consequence for metal directly: “metal roofs usually require a noncombustible roof deck or fire-resistant underlayment for the roof assembly to qualify as Class A.” The same guide notes that most asphalt shingle roofs carry a Class A rating without needing a noncombustible deck.

A covering does not have a fire class. A tested assembly does. Ask which listed assembly is being installed — deck, underlayment, and covering together — and whether a fire classification is required at your address at all.

Wind

Wind performance is a property of a specific assembly on a specific building. The model code ties asphalt shingle selection to a design wind speed and to test classifications rather than to a marketing number: shingles are classified under ASTM D7158 (classes D, G, and H) or ASTM D3161 (classes A, D, and F), and which class is acceptable depends on the design wind speed for the site. Exposure category, building height and geometry, pressure zone, enclosure, risk category, and the attachment schedule all enter the determination.

A number of miles per hour printed on a product wrapper is a test classification, not a code determination for your building. Two identical shingles on two identical houses can have different acceptable classifications because the sites differ.

Hail and impact

Impact classifications come from two different laboratory tests and both describe new product. The Insurance Institute for Business & Home Safety describes UL 2218 as a steel-ball test in which, “after two impacts in the same location, the product passes if no crack is visible on the back of the shingle and therefore is considered impact resistant to that size classification,” and FM 4473 as an ice-ball test in which “pure frozen water ice balls are shot or propelled at the shingle to achieve the appropriate kinetic energy for hailstones.” Class 4, the highest class in both, corresponds to a 2.00 inch impactor.

“Class 4 impact resistant” does not mean hail proof. IBHS states that these standards “evaluate new products and do not account for the effects of weathering, temperature, aging, or similar factors,” and neither test reproduces the irregular shape of real hail. An impact class is a comparison between products, not a promise about a storm.

Climate

The matrix separates climate fit from climate misfit because they are answered by different hazards. Freeze-thaw is the clearest example: the National Park Service notes that “poor quality porous tiles are particularly susceptible to breaking and exterior surface spalling during freeze-thaw cycles,” and that by letting in moisture, porous tiles can permit battens and structure to rot. The same tile in a hot-dry climate has none of that exposure.

Reflectivity is the second axis where fit and misfit diverge. The Department of Energy’s cool-roof guidance is explicit that “cool roofs achieve the greatest cooling savings in hot climates, but can increase energy costs in colder climates due to reduced beneficial wintertime heat gains.” A cool roof is a climate-conditional choice, not an upgrade.

Cool-roof benefit varies with climate, insulation, roof type, and mechanical system, and carries a possible heating tradeoff in cold climates. A high solar reflectance is not automatically better.

Code and jurisdiction

There is no nationwide building code for site-built construction in the United States. Every code figure on this page is model text, checked against the adopted code of a jurisdiction that publishes it officially and named here with its edition and effective date. The steep-slope figures are the 2024 International Residential Code, Chapter 9, adopted by Georgia as the 2024 IRC with Georgia Amendments effective 1 January 2026. The low-slope membrane figures are Section 1507 of the 2018 International Building Code as incorporated into the Minnesota State Building Code at Minnesota Rules part 1305.1507. The existing-building weight trigger is Section 3407.6.2 of the 2022 Oregon Structural Specialty Code, which Oregon superseded with its 2025 edition on 1 October 2025.

That spread is not sloppiness; it is the point. Three jurisdictions, three editions, three effective dates, and one of them already out of date — and none of them is necessarily yours. The same model sentence is law in some places, amended in others, superseded in others, and not adopted at all in others. The model codes even disagree with each other: wood shakes are 3:12 under the residential code and 4:12 under the commercial one.

Layer limits, ice-barrier extent, underlayment type, attachment schedules, and whether a re-cover is permitted at all are all decided by your adopted edition and its local amendments, not by a matrix.

Record the jurisdiction, the adopted edition, the amendments, and the effective date, and confirm with the authority having jurisdiction before treating any figure on this page as a requirement. A ZIP code is a routing hint, not a permitting-authority determination.

Access and site conditions

Everything the matrix asks of you is answerable without leaving the ground: slope from a photograph or a rafter measurement inside a finished room, material from a close-up, geometry from an aerial image, and structure from the drawings or from an engineer who visits. If a material decision seems to require someone to walk the roof, the person walking it should be a contractor with staging and fall protection, not you.

Do not climb onto a roof or into an attic to answer a question on this page. Ground-based, window-based, document-based, and installer-supplied evidence answers all of them.

Warranty and repair

Warranty length is the number most often confused with service lifeSection link

Nothing in the service-life column of the matrix is a warranty term, and nothing in a warranty document is a prediction of how long a roof lasts.

A warranty year and a service-life year are different units: A “50-year” shingle is a product sold under a document that runs fifty years, usually with a short non-prorated period followed by decades of declining, prorated material-only value. The federal capital-planning figure for asphalt shingle roofs, by contrast, is 20 years. Both numbers are real. They answer different questions, and only one of them is about how long the roof keeps water out.
Whose warranty, and on what: Manufacturer warranties cover the manufacturer’s product. Workmanship warranties cover the installer’s labour. System or enhanced warranties usually require that manufacturer’s own accessories and a credentialed installer, and carry registration deadlines and inspection obligations. Most roof failures are installation failures, which puts them in the second document rather than the first.
Transferability and proration change the comparison: Whether a warranty survives a sale, how many times, within what window, and for what fee is written differently in every document, and it is one of the few places where a material comparison and a resale conversation actually intersect. Read the transfer clause before treating a long warranty as a selling point.
What the matrix cannot show: A comparison table cannot tell you whether a warranty is worth anything, because that depends on the solvency of the issuer, the exclusions, and the law where you live. See how to read a roofing warranty for the clauses that decide it.

Repairability

Repairability is the column readers skip and regret. It has four components, and a material can score well on one and badly on the others.

Can one unit be replaced without disturbing others? Individually nailed, lapped units — slate, tile, wood, most shingles — can be, with the right tool. A welded membrane is patched rather than replaced. A standing seam panel usually runs eave to ridge, so replacing one panel means unclipping its neighbours.
Will the replacement match? Weathered granules, faded polymer, patinated metal, and quarried stone from a closed quarry all fail to match. This is why a partial repair on a fifteen-year-old roof is visible from the street and a partial repair on a slate roof often is not.
Will the unit still be made? Discontinued profiles are a real risk for tile and for proprietary composite products, and the National Park Service’s caution that artificial slates “may also have a tendency to fade over time” is a matching problem as much as an appearance one.
Who can do the work? The narrowest constraint of the four. Slate, historic clay tile, architectural metals, and torch-applied modified bitumen each need a trade that is not present in every market.

A warranty is a contract between a reader and whoever wrote it. What it covers, what voids it, whether it transfers, and how it is enforced are set by that document and by the law where the reader lives. Read the actual warranty for the product and the installer in front of you — not a summary of one, including this one.

Ask before you sign

Questions to ask an installerSection link

These are the questions that turn a material conversation into a scope conversation. None of them requires you to know roofing, and every one of them has a one-sentence answer from someone planning to do the work properly.

What is the measured slope of each plane on this roof, and does the material you are proposing meet the minimum for all of them?
The most common material mistake on a real house is a steep-slope covering carried onto a low-slope porch or addition. If the answer is a single number for the whole roof, ask about the porch.
Has anyone calculated whether the framing carries this material, and who signed that calculation?
For a change to tile, slate, or a heavier composite this is the gating question. “We do these all the time” is not a calculation. A named engineer is.
Which listed assembly are you installing — deck, underlayment, and covering — and what fire classification does the assembly carry?
This is the question that catches a covering being sold on a class the assembly does not have. A competent installer can name the listing and show you the sheet.
What wind classification is required at this address, and which class does this product carry?
The answer should reference the design wind speed for the site and the test class, not a marketing mph figure.
How many roofs in this material has your crew installed in the last three years, and can I see one that is at least five years old?
A five-year-old roof shows workmanship. A new one shows housekeeping. For slate, tile, standing seam, and torch-applied membrane this question matters more than any other on the list.
If a unit breaks in ten years, who repairs it, and will the same profile and colour still be available?
Discontinued profiles and thin repair trades are the two quiet ways a long-life material becomes a short-life roof. An honest installer will tell you which of the two applies.
What is the price per roofing square for each material you are proposing, with the same scope in each?
Two proposals are only comparable when they use the same unit and the same scope. Anything else is comparing a number to a different number.
Was this building built before 1990, and how are you handling the existing covering, felts, and mastics?
Older roofing products may contain asbestos. The right answer involves testing before disturbance by an accredited professional, not reassurance.

Require these in writing

The measured area and slope of every plane, and the material proposed for each.
The specific product, profile, and colour — not a material family.
The listed assembly by name, with deck, underlayment, and covering identified together.
The wind and, where relevant, impact classification the product carries, and the design wind speed for the site.
Any structural work the material choice triggers, priced separately and identified as such.
A deck-repair allowance with a unit rate and a documentation method.
Flashing, edge metal, and ventilation components by location, stated as new or reused.
A price per roofing square that can be compared with the other proposals on the table.
Who pulls the permit, and what the inspection will cover.

What goes wrong

Misconceptions and failure modesSection link

Common misconceptions

Common belief
A 50-year shingle lasts 50 years.
What is actually true
Fifty is the length of a warranty document, not a prediction. The federal capital-planning figure for an asphalt shingle roof is 20 years, and observed lives vary widely with climate, ventilation, colour, and installation quality. Service life is always a planning range with a stated basis, never a warranty term. See why “30-year shingle” is a warranty term.
Common belief
Metal roofs are lighter, so weight is never an issue with metal.
What is actually true
Metal is usually the lightest of the durable coverings, and that is genuinely why it is often the answer where structure is marginal. But weight is a product-specific number, not a family property: stone-coated steel, heavy-gauge copper, and zinc are all “metal” and none of them weighs the same. The number you need is on the product data sheet, not in a category.
Common belief
Class 4 shingles mean no more hail claims.
What is actually true
Class 4 is the top class of a laboratory test on new product. IBHS is explicit that the standards “evaluate new products and do not account for the effects of weathering, temperature, aging, or similar factors,” and neither the steel ball nor the ice ball reproduces the irregular shape of real hail. An impact class compares products; it does not describe a storm.
Common belief
A Class A material makes a Class A roof.
What is actually true
Classification applies to a tested assembly. The model code says so in its own words, and its own exceptions depend on the deck: metal, tile, and slate are deemed Class A on noncombustible decks, and metal on a combustible deck usually needs a fire-resistant underlayment for the assembly to qualify. The covering is one component of the answer.
Common belief
Synthetic slate is the same decision as slate, just cheaper.
What is actually true
It is a different decision. The National Park Service observes that because labour, flashings, and removal costs are equal, “the total initial cost of an artificial slate roof is only marginally less than a natural slate roof,” and that over the long term “several artificial slate roofs will have to be installed during the life span of one natural slate roof.” The trade is a lighter, cheaper, more available product against a shorter and much less documented record.
Common belief
The right material is the one with the best warranty.
What is actually true
Warranty length correlates with marketing budget more reliably than with performance. What decides the outcome is whether the assembly suits the slope, whether the structure carries it, whether the climate is a fit, whether the flashing details are right, and whether it can be repaired. All five are settled before the warranty is read.
Common belief
Low-slope roofs are just flat roofs and are always a problem.
What is actually true
Below 2:12 the roof stops shedding and starts holding out water with a continuous membrane. That is a different, well-understood technology, not a defective version of a steep roof. What causes the reputation is putting a shed-water material on a hold-water slope. See low-slope roofing on houses.

How it actually fails

Steep-slope covering carried onto a low-slope plane: A porch, dormer, or addition below the covering’s minimum slope gets the same shingles as the main roof because it is easier to quote one material. Water no longer clears the laps, and moves sideways under them.What you can see: Recurring leaks confined to the low plane, often only in prolonged rain or snowmelt. From the ground, a visibly shallower section with the same covering as the steep section.
Heavy material added without a structural check: Tile, slate, or a second layer is added to framing sized for something lighter. The load is permanent, and the deflection develops slowly enough that nobody connects it to the re-roof.What you can see: A ridge line that has developed a sag, doors and windows on the top floor that start binding, or new cracking at the top of interior partitions. Any of these is a call to an engineer, not to a roofer.
Material chosen for the hazard the owner was thinking about: A hail-resistant covering is specified in a market whose real exposure is wind-driven rain, or a reflective membrane is specified in a heating-dominated climate. The chosen property is real; it is simply not the property that governs at that address.What you can see: A roof that performs adequately and delivers none of the benefit it was bought for. Usually invisible until an energy bill or a storm makes it visible.
Long-life material with no local repair trade: Slate, historic tile, or an architectural metal is installed in a market with no one to maintain it. Ordinary damage that should be a two-hour repair becomes a deferral, then a leak, then a replacement decades early.What you can see: Repairs done in the wrong material — asphalt patches on a slate roof, sealant across a tile field, mismatched replacement units. Each one is a record that nobody qualified was available.
Discontinued profile: A proprietary tile or composite profile is withdrawn, and no matching unit exists for a repair. The roof is intact and unrepairable at the same time.What you can see: Visible patches in a near-match colour, or an installer proposing to rob units from a rear slope to repair a front one.
Comparison made on price per project rather than per square: Three proposals name three materials, three scopes, and three totals. Without a common unit and a common scope, the cheapest total is simply the smallest promise.What you can see: A proposal with no area stated, no scope list, and a single number at the bottom. See how to make three quotes comparable.

Sources and further readingSection link

Understanding Roofing

Scope and limitations

It publishes no installed cost figure for any material.
This site holds no per-material cost dataset that meets its own evidence standard, and its cost methodology page states in its own words that only the asphalt shingle rate behind its beta estimator is traceable to a recorded source, that the other five material rates are unattributed beta values, and that they should not be quoted.
An honest blank is the correct entry until a dataset exists.
It publishes a weight figure for exactly one material family — concrete tile — and only from one manufacturer's currently published technical page.
Roofing weight is genuinely product-specific, published weight figures circulating online disagree by factors of two, and no government or standards source giving weights for every covering family could be found and verified.
It cannot tell you whether your framing carries a heavier covering.
That is an arithmetic question about your building, answered by a licensed structural engineer.
The 5 percent gravity-load trigger quoted here comes from a code edition Oregon has already superseded, and the model text behind it has been written more than one way; it is a worked example of a screening rule, not a national permission and not a current requirement anywhere on this page's authority.
It cannot tell you what your jurisdiction requires.
Every code figure here is model text, checked against three official adoptions with their editions and effective dates recorded — the 2024 IRC with Georgia Amendments (effective 1 January 2026), the 2018 IBC as incorporated into the Minnesota State Building Code at Minnesota Rules 1305, and the 2022 Oregon Structural Specialty Code (superseded 1 October 2025).
Your adopted edition, its amendments, its effective date, and your authority having jurisdiction govern, and none of the three is likely to be yours.
It quotes section wording that was read on UpCodes, a commercial code aggregator.
UpCodes is not a jurisdiction and is not the official publisher of anyone's adopted code; where this page relies on a code figure, an official adoption or an independent federal or trade source is cited alongside it, and the UpCodes entries are recorded only as the copy actually read.
The authoritative model text is published by the International Code Council, and the authoritative adopted text by the jurisdiction itself.
Its minimum slopes are new-construction minimums.
For replacement and re-cover of an existing low-slope roof, the model codes waive the 1/4:12 figure in favour of a positive-drainage performance requirement, so a re-roof is judged on whether the roof drains rather than on the number in the diagram.
The service-life column rests on a single federal table built for multifamily capital needs assessment.
It is a planning constant chosen so that a reserve is funded in time, not a prediction for one house, and HUD's own instructions allow an assessor to override it with an assessed remaining useful life.
It has no row for composite or synthetic slate in the federal service-life table, because that table has none.
The product category is young, and no source found offers a defensible multi-decade figure for it.
It cannot rank materials.
Every row wins on some column and loses on another, and which column governs is a fact about your building and your market, not about roofing.

Current State Minimum Codes for Construction
Georgia Department of Community Affairs
That Georgia's mandatory state minimum residential code is the International Residential Code, 2024 Edition, with Georgia Amendments — the jurisdiction, edition, and adopting authority behind every steep-slope figure on this page.
It records what Georgia has adopted. It does not reproduce the section text, it does not apply outside Georgia, and Georgia's own amendment packet modifies parts of the model code.
New Codes Jan 2026 — Board adoption of the 2024 International Residential Code with Georgia Amendments
Georgia Department of Community Affairs / 9 December 2025
The effective date of the Georgia adoption relied on here: the 2024 International Residential Code with Georgia Amendments took effect 1 January 2026.
An adoption announcement. It states the edition and the effective date; it does not reproduce any code section, and it binds nothing outside Georgia.
2024 International Residential Code, Chapter 9: Roof Assemblies — R902.1, R905.2.2, R905.3.1, R905.3.2, R905.4.2, R905.5.2, R905.6.1, R905.6.2, R905.7.2, R905.8.2
UpCodes — a commercial code aggregator, reproducing International Code Council model text in the Georgia 2024 IRC view / 2024 edition
The wording read for this page: that Class A, B and C classifications apply to roof assemblies tested to ASTM E108 or UL 790, and the four deck-dependent exceptions; minimum slopes for asphalt shingles (2:12), clay and concrete tile (2 1/2:12), metal roof shingles (3:12), mineral-surfaced roll roofing (1:12), slate (4:12), wood shingles (3:12) and wood shakes (3:12); and the solid-sheathing requirement for tile and slate.
UpCodes is a commercial publisher, not a jurisdiction. This is a convenience reproduction and is not the official adopted text of Georgia or anywhere else; the adoption itself is recorded by the Georgia DCA sources above, and the International Code Council publishes the authoritative model version at codes.iccsafe.org. It is cited here only as the copy actually read, never as the authority for a requirement.
2024 International Residential Code, Chapter 9 — R905.2.2 and R905.2.4.1 (asphalt shingle wind resistance)
UpCodes — a commercial code aggregator, reproducing International Code Council model text in the General Services Administration 2024 IRC view / 2024 edition
The wording read for this page: that asphalt shingles are classified under ASTM D7158 (classes D, G, H) or ASTM D3161 (classes A, D, F), and that which classification is acceptable depends on the design wind speed for the site. The same relationship is stated independently by the Department of Energy guide cited below, which is what this page actually rests the claim on.
A commercial reproduction of model text, not adopted law and not a jurisdiction's official publication. It is not a wind-design determination for any building; that is site- and building-specific engineering.
Metal roof panels — deck slope, Section 1507.4.2
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Texas Windstorm Insurance Association 2024 view
The wording read for this page for the three metal-panel minimums: 3:12 for lapped, non-soldered seams without applied lap sealant; 1/2:12 with applied lap sealant; and 1/4:12 for standing-seam systems. All three are stated independently by the Department of Energy metal roofs guide cited below, which is what this page rests them on.
A commercial reproduction, read in a windstorm-insurance eligibility code rather than a general building code. TWIA's building code governs insurability in the Texas seacoast territory; it is not a statewide building code. Section numbering and the adopted edition differ between jurisdictions and between the residential and commercial codes.
Wood shakes — deck slope, Section 1507.9.2
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Texas Windstorm Insurance Association 2024 view
That the commercial model code sets wood shakes at not less than 4 units vertical in 12 units horizontal — the figure this page names only to show that it disagrees with the residential model code, which sets 3:12 at IRC R905.8.2.
A commercial reproduction of a windstorm-insurance eligibility code, not adopted general law. This page does not use the 4:12 figure for its matrix, because its subject is houses and the residential model code governs them. Wood shake and wood shingle minimums have differed between code families and between editions; check your adopted edition.
Minnesota Rules, part 1305.1507 — Section 1507, Requirements for Roof Coverings
Office of the Revisor of Statutes, State of Minnesota (official publisher of the Minnesota Administrative Rules) / Rule text electronically published 19 June 2015
The low-slope minimums in the diagram and matrix, in the text of an adopted state building code rather than a commercial reproduction: a minimum of one-fourth unit vertical in 12 units horizontal (2-percent slope) for built-up, modified bitumen, thermoset single-ply and thermoplastic single-ply roofs, with an exception of one-eighth unit vertical in 12 units horizontal (1-percent slope) for coal-tar built-up roofs.
Adopted law in Minnesota only, and Minnesota amends the model text — the rule adds an exception for roofs designed for water accumulation under Section 1611.2. It says nothing about what any other jurisdiction has adopted.
Minnesota Rules, part 1305.0011 — incorporation of the International Building Code
Office of the Revisor of Statutes, State of Minnesota / Published 31 March 2020
Which model edition the Minnesota rule above reproduces: “IBC means the 2018 edition of the International Building Code as promulgated by the International Code Council, Inc.”, incorporated by reference into the Minnesota State Building Code.
It fixes the edition for Minnesota only. Minnesota's incorporated edition changes over time, and no other state's edition can be inferred from it.
Roof slope guidelines
Mark S. Graham, National Roofing Contractors Association — Professional Roofing / 1 August 2018
That the model code prescribes a 1/4:12 minimum slope for asphalt built-up, polymer-modified bitumen, thermoset and thermoplastic single-ply and liquid-applied membranes in new construction; and the qualification this page now carries — that for replacement and re-cover of existing low-slope roof systems the model codes waive the 1/4:12 minimum and apply a performance-based positive-drainage requirement instead.
Trade-association technical guidance written against the 2018 editions, not adopted law. NRCA's own recommendations are performance-based rather than a prescribed minimum slope, and it says so.
Built-up roofs — slope, Section 1507.10.1
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Illinois Building Code 2021 view
The wording read for this page: a design slope of not less than 1/4:12 for drainage on built-up roofs, and not less than 1/8:12 for coal-tar built-up roofs. The same figures are in the Minnesota rule above, which is what this page rests them on.
A commercial reproduction, not an official jurisdiction publication. The Illinois view is also narrower than it looks: Illinois has no statewide building code for most private construction.
Modified bitumen roofing — slope, Section 1507.11.1
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Illinois Building Code 2021 view
The wording read for this page: a design slope of not less than 1/4 unit vertical in 12 units horizontal for drainage on modified bitumen roofing. Corroborated by the Minnesota rule and the NRCA guidance above.
A commercial reproduction, not an official jurisdiction publication.
Thermoset single-ply roofing — slope, Section 1507.12.1
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Texas Windstorm Insurance Association 2018 view
The wording read for this page: a design slope of not less than 1/4 unit vertical in 12 units horizontal for drainage on thermoset single-ply membrane roofs — the EPDM row of the matrix. Corroborated by the Minnesota rule and the NRCA guidance above.
A commercial reproduction of a windstorm-insurance eligibility code, not adopted general law. Confirm the edition adopted where the building stands.
Thermoplastic single-ply roofing — slope, Section 1507.13.1
UpCodes — a commercial code aggregator, reproducing International Building Code model text in the Texas Windstorm Insurance Association 2018 view
The wording read for this page: a design slope of not less than 1/4 unit vertical in 12 units horizontal for thermoplastic single-ply membrane roofs — the TPO and PVC rows of the matrix. Corroborated by the Minnesota rule and the NRCA guidance above.
A commercial reproduction of a windstorm-insurance eligibility code, not adopted general law. Confirm the edition adopted where the building stands.
Oregon Structural Specialty Code adoption
Building Codes Division, State of Oregon (the adopting authority)
The currency statement this page now carries: Oregon's adopted structural code is the 2025 Oregon Structural Specialty Code, based on the 2024 International Building Code, Fire Code and Existing Building Code, effective 1 October 2025 with a six-month phase-in and mandatory compliance from 1 April 2026. The 2022 edition quoted on this page is therefore the previous edition, not current Oregon law.
It records Oregon's adoption and effective dates. It does not reproduce Section 3407.6.2, and it says nothing about any other state.
Addition or replacement of roofing — Oregon Structural Specialty Code 2022, Section 3407.6.2
UpCodes — a commercial code aggregator, reproducing the 2022 Oregon Structural Specialty Code / 2022 edition
The wording quoted on this page: that an existing gravity load-carrying element whose design gravity load rises by more than 5 percent must be replaced or altered to carry new-construction loads, and the exception for a second layer of roof covering weighing 3 pounds per square foot or less over a single existing layer.
A commercial reproduction of an edition Oregon has since superseded — the 2025 OSSC took effect 1 October 2025, per the Building Codes Division source above. Quoted here as a worked example of the shape such a trigger takes, not as a requirement in force anywhere. Elsewhere the equivalent provision may be a different edition, differently amended, or absent, and it is never a substitute for an engineer's calculation on a specific building.
CNA e-Tool Estimated Useful Life Table (numbering by ASTM 2018-08 outline)
U.S. Department of Housing and Urban Development
The service-life column and the sixty-year worked example: asphalt shingle 20 years, metal 50, slate shingle 75, clay/cementitious barrel tile 60, wood shingle and cedar shakes 25, low-slope built-up with gravel finish 20, low-slope built-up without mineral or gravel finish 10, adhered rubber membrane (EPDM) 15, thermoplastic membrane (TPO, vinyl) 15, and rubberized/elastomeric white or cool roof 15.
Built for capital needs assessments on multifamily properties, not for predicting the life of one house's roof. HUD states the standard figure is fixed but that an assessor may enter a different assessed remaining useful life with justification. The table has no row for composite or synthetic slate, and no row for PVC separate from thermoplastic.
Asphalt Shingle Roofs
U.S. Department of Energy, Building America Solution Center (PNNL)
That asphalt shingles should only be installed at 2:12 or greater, with double underlayment from 2:12 up to 4:12; the ASTM D3161 and D7158 wind-class-to-design-wind-speed relationship; that impact-rated shingles are classified under UL 2218 Class 1 through 4; and that most asphalt shingle roofs carry a Class A fire rating without requiring a noncombustible deck.
Best-practice guidance for builders, not adopted law. It does not publish a service-life figure or a weight for asphalt shingles.
Metal Roofs
U.S. Department of Energy, Building America Solution Center (PNNL)
That a metal roof can last up to 50 years or more; the metal-panel and metal-shingle slope minimums; that metal roofs usually require a noncombustible deck or fire-resistant underlayment for the assembly to qualify as Class A; the SRI targets of 78 for low-slope and 29 for steep-slope; the use of stainless fasteners or clips within 3,000 feet of a saltwater coastline; and that metal roofing tends to cost more than asphalt shingles.
Guidance rather than code, and it does not publish weights. Its coastal-fastener figure is a best-practice threshold, not a universal requirement.
Low-Slope (Flat) Roofs
U.S. Department of Energy, Building America Solution Center (PNNL)
That low-slope roof assemblies are those with a slope less than 2:12, and that the residential low-slope family covers built-up bitumen, modified bitumen, and single-ply membranes including TPO, PVC, KEE and EPDM.
It does not publish service-life figures, weights, or costs for these systems.
Preservation Brief 29: The Repair, Replacement, and Maintenance of Historic Slate Roofs
U.S. National Park Service / 1992
That properly installed slate roofs last 60 to 125 years or longer and that some have lasted over 200 years; the quarry-specific figures of about 125 years for Vermont and New York slates, 175 years or more for Buckingham Virginia, in excess of 60 years for Pennsylvania Soft-Vein, and roughly 100 and at least 200 years for Pennsylvania Hard-Vein and Peach Bottom respectively; that these life spans should be used only as a general guide; that repairs damage additional slates even in expert hands; and that roof timbers showing deflection, cracking or twisting call for a structural engineer experienced with older buildings.
Written for historic buildings and for slate specifically. It is not a code determination anywhere, and its quarry-specific figures describe historic American slates rather than every slate on the market today.
Preservation Brief 30: The Preservation and Repair of Historic Clay Tile Roofs
U.S. National Park Service / 1992
That clay tiles are heavy and that it is important the roof structure be sound; that poor-quality porous tiles are particularly susceptible to breaking and surface spalling during freeze-thaw cycles and can let moisture reach battens and structure; and that wood shingles were historically preferred over tile in part because they were much lighter and required less heavy and less expensive roof framing.
Historic-buildings guidance. It publishes no weight in pounds per square foot and no service-life figure.
Should I Replace My Slate Roof with a Synthetic?
U.S. National Park Service
That artificial mineral-fiber slate is not recommended for restoration work and may tend to fade over time; that the total initial cost of an artificial slate roof is only marginally less than a natural slate roof because labour, flashings and removal costs are equal; and that over the long term several artificial slate roofs will have to be installed during the life span of one natural slate roof.
Written for historic properties and for the preservation decision. It is not a product evaluation of the modern polymer composite category, and it publishes no service-life figure for synthetics.
Purchasing Energy-Efficient Cool Roof Products
U.S. Department of Energy, Federal Energy Management Program
That cool roofs achieve the greatest cooling savings in hot climates but can increase energy costs in colder climates due to reduced beneficial wintertime heat gains, and that high thermal emittance as well as high solar reflectance matters.
Federal procurement guidance. It is not a determination about any particular building's energy balance.
Roof 101 — impact, wind and fire testing of roof coverings
Insurance Institute for Business & Home Safety
The description of UL 2218 as a steel-ball test passed when no crack is visible on the back of the shingle after two impacts in the same location; FM 4473 as an ice-ball test using pure frozen water ice balls; Class 4 corresponding to a 2.00 inch impactor in both; and that these standards evaluate new products and do not account for the effects of weathering, temperature, aging, or similar factors.
Written primarily around asphalt shingles. It does not establish what any particular roof will do in a particular hailstorm, and it is not an insurance-coverage determination.
Structural Considerations for Reroofing Projects
Wiss, Janney, Elstner Associates, Inc.
That Section 706.2 of the 2018 International Existing Building Code is triggered by an increase in design dead, live or snow load of more than 5 percent; that editions before 2018 instead counted additional dead loads increasing forces in structural elements by more than 5 percent; and that the exception covers a second layer of roof covering weighing 3 pounds per square foot or less.
Engineering guidance, not adopted law and not a calculation for any specific building. Its citations are explicitly to the model 2018 IEBC — the document states that local adopted provisions may differ and that the reader must check with the authority having jurisdiction. It does not analyse the 2021 or 2024 editions of Section 706.2.
Technical Specifications — approximate weight per 100 square feet by tile profile
Eagle Roofing Products (manufacturer, product-specific)
The one weight figure this page publishes: standard-weight concrete roof tile profiles at approximately 900 to 970 pounds per 100 square feet, and lightweight profiles at approximately 550 to 720 pounds per 100 square feet, with availability varying by region.
One manufacturer's published figures for its own products, at named profiles and named plants. They are not a weight for concrete tile generally, and they are not a weight for clay tile at all. Use the data sheet for the product actually being proposed.
Fall Protection in Residential Construction
U.S. Occupational Safety and Health Administration
That falls are the leading cause of death for workers engaged in residential construction, and that workers in residential construction six feet or more above lower levels must be protected by conventional fall protection.
An occupational-safety standard for employers and workers. It is not homeowner guidance; the fact that trained workers use fall protection is a reason for an untrained reader to stay off the roof, not a procedure to copy.
How do I know if I have asbestos in my home?
U.S. Environmental Protection Agency
That shingles and similar home products may contain asbestos; that EPA only recommends testing suspect materials if they are damaged or if planned work would disturb them; and that samples should be taken by a properly trained and accredited asbestos professional.
General homeowner guidance. It does not identify which specific roofing products contain asbestos, and state and local rules on testing, notification, and disposal vary.

Suggest a correction