For homeowners, property buyers, and preservation and design professionals

On a slate roof, the fasteners and the flashings run out before the slate does.

Steep-slope roofs · single-family, institutional, and historic buildings

Quarried stone measured in centuries is hung on nails and metal measured in decades. Which of those runs out first is the entire planning problem, and the stone is rarely the answer.

30-second answer

How long does a natural slate roof last, and what actually ends it?

Properly installed slate roofs last 60 to 125 years or more, and some slate has outlived 200. The nails and flashings holding it up do not. The Park Service records that galvanized and steel nails usually rust out long before the slate deteriorates, and that metal flashings often wear out before it. Specify those two, not the stone.

Stop — immediate hazard

Slates that are slipping out of position can fall. Treat the ground under them as a hazard before you treat the roof as a project.

A single slate is a slab of stone with a sharp edge, and it arrives at the ground edge-first from roof height. Corroding fasteners do not fail one at a time in a tidy sequence: as Historic Environment Scotland puts it, as more and more nails decay, numerous slates may become loose and slip out of place.

Do not go up to look. Slate is brittle and slippery, your weight breaks slates that were sound, and the National Park Service is direct about it: roof surveys should be carried out from the ground using binoculars, or from a lift, not by walking the roof. Falls are the leading cause of death for workers in residential construction, and those are trained people wearing fall protection.

Keep people, pets, parked vehicles, and any doorway or walkway directly below a slipping roof clear until a contractor who works in slate has looked at it. If a ceiling below is sagging, bulging, or holding water, stay out from under it and leave the room.

Learning paths and saved lessons

At a glance

The short versionSection link

Every code dimension below is from the 2024 International Residential Code, which is model text — language a state or local government may adopt, amend, or decline. It is not the law where you live until your jurisdiction adopts it. Confirm the adopted edition and its amendments with your authority having jurisdiction before treating any number here as a requirement.

What it is: Metamorphic stone, split by hand along its cleavage planes into flat unitsNot manufactured to a formula. Composition, strength, and durability vary between quarries and between beds inside one quarry, because the degree of metamorphism was never uniform.
Minimum slope (model code): 4:12 and steeper2024 IRC R905.6.2. Trade installation guidance says the same thing more bluntly: below 4:12 is not recommended.
Headlap (model code): 4 in · 3 in · 2 in, by slope2024 IRC Table R905.6.6: 4 in from 4:12 up to 8:12, 3 in from 8:12 up to 20:12, 2 in at 20:12 and steeper. Headlap, not the visible surface, is what makes the roof watertight.
Fasteners (model code): Two per slate2024 IRC R905.6.6. Trade guidance adds the metal: solid copper or Type 304 stainless, smooth shank, and no nail guns. This is the component the roof usually fails at.
Material standard: ASTM C406, grades S1 / S2 / S32024 IRC R905.6.4 requires slate shingles to comply with ASTM C406. The grade is a laboratory result on sampled stone, not a statement about a finished roof.
Service-life planning range — the stone: 60 to 125 years or longerNational Park Service, Preservation Brief 29. Some slates have lasted over 200 years. This is a planning range with a stated basis, not a warranty term and not a promise about any specific roof.
Service-life planning range — the metal: Shorter, and it is what ends the roofThe Park Service states it plainly: because slate is so durable, metal flashings often wear out before the slate does. Plain steel and galvanized nails will usually rust out long before the slate begins to deteriorate.
Installed weight: Roughly 6 to 17 lb per sq ft at common thicknessesDerived below from a published stone-density band and code headlap, not quoted from a marketing sheet; a denser slate lands above the band. Model rafter span tables are published for dead loads of 10 psf and 20 psf, which is why a change to slate is a structural question.
Roofing over it: Prohibited under widely adopted model reroofing textModel text prohibits a roof recover where the existing covering is slate, clay, cement, or asbestos-cement tile. Where a jurisdiction has adopted it, a slate roof comes off before a new roof goes on — confirm your adopted edition rather than assuming either way.

ASTM C406 grades for roofing slate, and what a grade is and is not a claim about.
Grade	Expected service life associated with the grade	What the grade is a claim about	What it cannot tell you
S-1	Over 75 years	A sample of quarried stone that met the standard's thresholds for water absorption, breaking load, and depth of softening in a laboratory. Trade guidance published in IIBEC Interface is that only S-1 should be used for roofing.	Whether the slate on the pallet delivered to your building is the slate that was sampled, and whether the roof it is laid on will be nailed and flashed well enough to reach that number.
S-2	40 to 75 years	The same three tests at lower thresholds. In practice this is a covering expected to be replaced within a normal ownership lifetime rather than handed on.	How the lifecycle argument for slate survives the change. Slate earns its cost by not being replaced; a 40-year slate roof is a very expensive way to buy 40 years.
S-3	20 to 40 years	The lowest tier the standard recognises. Stone that met the specification, at the bottom of it.	Why you would take it. On service life alone this tier overlaps the range published for asphalt shingles — without asphalt's price, its weight advantage, or its supply of installers.

Read this table one item at a time

S-1

Expected service life associated with the grade: Over 75 years
What the grade is a claim about: A sample of quarried stone that met the standard's thresholds for water absorption, breaking load, and depth of softening in a laboratory. Trade guidance published in IIBEC Interface is that only S-1 should be used for roofing.
What it cannot tell you: Whether the slate on the pallet delivered to your building is the slate that was sampled, and whether the roof it is laid on will be nailed and flashed well enough to reach that number.

S-2

Expected service life associated with the grade: 40 to 75 years
What the grade is a claim about: The same three tests at lower thresholds. In practice this is a covering expected to be replaced within a normal ownership lifetime rather than handed on.
What it cannot tell you: How the lifecycle argument for slate survives the change. Slate earns its cost by not being replaced; a 40-year slate roof is a very expensive way to buy 40 years.

S-3

Expected service life associated with the grade: 20 to 40 years
What the grade is a claim about: The lowest tier the standard recognises. Stone that met the specification, at the bottom of it.
What it cannot tell you: Why you would take it. On service life alone this tier overlaps the range published for asphalt shingles — without asphalt's price, its weight advantage, or its supply of installers.

ASTM C406/C406M covers the material characteristics, physical requirements, and sampling appropriate to selecting slate for use as roof shingles, and classifies it as Grade S1, S2 or S3 against breaking load, absorption, and depth of softening. The underlying test methods are ASTM C120 (flexure), C121 (water absorption), and C217 (weather resistance). Since 2005 the flexure requirement has been a minimum breaking load of 575 lbf rather than the 9,000 psi modulus of rupture required before. The service-life figures in the second column are the ones the trade literature associates with each grade; they are not printed on a pallet and they are not a warranty. The standard contains no test for oxidation, which is why slate can comply with it and still develop rust staining after installation.

Tradeoffs

This page's advice — buy the fastener and the flashing, not the stone — and where that advice is wrongSection link

The position taken here is that on a slate roof the specification decisions that matter most are the ones nobody looks at: nail metal, flashing metal, and whether the structure and the labour market can carry the roof for a century. That framing is not right for every building.

Best when

The building already has a slate roof and the framing has carried it for decades — the structural question is answered by the building's own history rather than by an assumption.
The failure you are looking at is slates sliding out of position while the slates themselves ring solid. That is a fastener problem, and salvaging and re-laying the existing slate is often the correct answer.
The ownership horizon is long or institutional — a church, a school, a courthouse, a family property expected to stay in the family. Slate's whole economic argument is the replacement that does not happen.
A slater with real slate work behind them is reachable and will still be reachable in twenty years, because a slate roof needs a competent repair every few years far more than it needs a replacement.
The roof is steep. The Park Service notes that the steeper the pitch, the longer the slate can be expected to last, and that spires and Mansard slopes often keep their original slate long after flatter areas have been replaced.
The building is historic, and the pattern, coursing, and detailing of the roof are part of what makes it that building.

Think twice if

No engineer has looked at the framing and the building has never carried slate. A conversion to slate is a structural project with a roofing component, not the reverse, and no table on this page settles it.
You are planning to sell within a decade or two. The lifecycle case for slate is built entirely on the century you will not own.
The nearest contractor with genuine slate experience is hours away. A roof whose repairs cannot be bought locally will be repaired badly, and bad slate repairs — mastic, face-nailing, mismatched units — shorten it faster than weather does.
The proposal quotes a grade and a colour but is silent on nail metal, flashing metal, and flashing weight. Those three lines decide when the roof ends.
The roof has a lot of low-slope area, long valleys, or built-in gutters. Those are the parts that fail first, they are metal, and on a slate roof they are the expensive kind of metal.
The house was built before 1990 and the existing covering is a slate look-alike rather than slate. Asbestos-cement shingles were sold as artificial slate for decades, and the question becomes a testing question before it is a roofing one.

What changes the answer

Whether the framing already carries this weight. A re-slate of an existing slate roof and a conversion from asphalt to slate are different projects with different risk.
The specific slate and its origin. The Park Service reports service lives from 'in excess of 60 years' to 'at least 200 years' across U.S. quarry districts — a spread wider than the difference between whole material categories.
The nail metal. Copper or stainless keeps the fastener out of the failure sequence; plain steel or galvanized puts it in.
The flashing metal and its weight, and whether one metal is used throughout so that galvanic corrosion is not designed into the roof on day one.
Slope, and how much of the roof is at low slope, in a valley, or under a concentrated flow — those areas deteriorate faster than the field.
Your climate's freeze-thaw and ice-damming behaviour, which acts on the parts of the roof that are already the weakest.
Whether anyone will be able to buy a matching slate in thirty years. Ribbon slate is no longer manufactured; whole quarry districts have closed.

The mechanism

The stone does not keep the water out. The overlap does.Section link

Understand this one drawing and the rest of the page follows from it, including why the fastener is the part you will never inspect.

A section through four courses at a 3 in headlap on a 20 in slate. The full text description is available to screen readers and every dimension is repeated in the arithmetic below. Schematic: the thickness of each course is exaggerated so the layers read apart, and it is not a construction detail.Original diagram, Understanding Roofing.

A slate roof is not waterproof and was never built to be. It is water-shedding: each course of stone laps over the course below it, and the redundancy that matters is the headlap — the vertical overlap between a course and the course two below it. Water that gets between two slates in the same course meets the slate underneath, then has to travel up and over the headlap before it can reach the deck. That is the whole waterproofing strategy, and it is why the model code dimensions headlap by slope rather than dimensioning the slate.

The 2024 IRC gives 4 inches of headlap from 4:12 up to 8:12, 3 inches from 8:12 up to 20:12, and 2 inches at 20:12 and steeper. A flatter roof gets more overlap because water moves across it more slowly and wind has longer to push it uphill. The Park Service describes exactly the same historic practice, including the 4:12 floor below which a slate roof was not laid.

Headlap decides exposure, and exposure decides everything else

The visible band of each course — its exposure — is not a free choice. Once the slate length and the headlap are fixed, the exposure is fixed with them, because each slate has to reach back far enough to lap the course two below. The Park Service gives the relationship: exposure is the slate length minus the headlap, divided by two.

exposure = (slate length − headlap) ÷ 2

That single line is what makes a slate roof heavy, expensive, and long-lived at the same time. Getting 8½ inches of visible roof out of a 20-inch slate means laying about two and a third square feet of stone for every square foot of roof you can see.

The nail hangs the slate. It does not clamp it.

Slate is fastened differently from every other steep-slope covering, and the difference is the source of most of the damage done to slate roofs by people who have not worked on one. The Park Service is explicit: slate nails should not be driven tight as is the case with asphalt and wood shingles, but set so the slate is permitted to hang freely on the nail shank. Drive the nail too far and it cracks the slate. Leave it projecting and it punches a hole through the slate lying over it. The Slate Roofing Contractors Association guidance says the same in specification language — nails are driven so the head lies within the countersunk hole and does not rub the overlying slate — and adds that pneumatic and electric nail guns are not permitted for slate.

Two nails per slate is the model-code requirement, placed above the head of the slate below and as far from the centre of the slate as practical. Look at the drawing again and notice where that puts them: under the course above, over the head of the course below, buried in stone. From the day the roof is finished, nobody sees a nail again — not from the ground, not from a lift, not from the attic, not in a photograph, not in an infrared scan.

So the roof often fails at the one component nobody can inspect

The Park Service records the consequence in a single sentence: Non-ferrous slater’s nails, such as solid copper or stainless steel, should always be used since plain steel and galvanized nails will usually rust out long before the slate itself begins to deteriorate. And it names the result: The rusting of nineteenth century cut nails is a common cause of slate loss on historic roofs.

The British trade name for this condition is nail sickness. Historic Environment Scotland defines it exactly: ‘Nail sickness’ occurs when the nails rust through, and causes individual slates to slip, and as more and more nails decay, numerous slates may become loose and slip out of place. The stone is fine. The roof is over.

The same logic applies one layer out, to the metal. The Park Service puts it as a design instruction — flashings are the weakest point in any roof, so given the permanence of slate, it is poor economy to use anything but the most durable of metals and the best workmanship for installing flashings — and then as an observation about roofs that are already built: because slate is so durable, metal flashings often wear out before the slate does. Everything on the flashing page applies here with the clock turned up: a valley lining that would see out an asphalt roof measured in decades is a scheduled failure under a slate roof measured in generations.

Worked example

The arithmetic of one square of slateSection link

Everything people find surprising about slate — the weight, the labour, the number of concealed fasteners — falls out of two published numbers and one division. Here is the whole calculation, with its inputs, so you can check it.

Take an ordinary size: a slate 10 inches wide by 20 inches long, laid at the 3-inch headlap the 2024 IRC gives for slopes from 8:12 up to 20:12.

Step 1 — exposure

exposure = (20 in − 3 in) ÷ 2 = 8.5 in

Eight and a half inches of a twenty-inch stone is all you will ever see. The other 11½ inches is buried, doing the work.

Step 2 — how many slates cover 100 square feet

Each slate covers its width by its exposure: 10 in × 8.5 in = 85 sq in = 0.590 sq ft.

100 sq ft ÷ 0.590 sq ft = 169.4 → about 170 slates per square

The published exposure-and-headlap table in the Slate Roofing Contractors Association guidelines gives, for a 10 by 20 in slate at a 3 in headlap, an exposure of 8½ in and 170 slates per square. The arithmetic and the trade table agree, which is the point of showing both.

Step 3 — how many fasteners that is

The model code requires two fasteners per slate. So one roofing square carries about 340 nails. A modest 25-square roof carries roughly 4,250 slates and 8,500 nails — and, as the diagram above shows, not one of those nails is visible again after the course above it is laid.

That is the sentence this whole page is built on. The component with the shortest life on a slate roof is also the component with zero inspectability. You cannot check it, so you have to specify it — which is why “what nail are you using” is the first question in the list further down.

Step 4 — how much stone is actually up there

Every square foot of visible roof needs length ÷ exposure = 20 ÷ 8.5 = 2.35 square feet of stone. So a square of finished roof — 100 sq ft — contains about 235 square feet of slate. That overlap is the waterproofing, and it is also the weight.

Step 5 — the weight, derived

The U.S. Geological Survey’s 1914 survey of American slate publishes a laboratory series on Arkansas slates with weights of 159 to 176 pounds per cubic foot, at apparent specific gravities of about 2.54 to 2.82. That is the band used below. Treat it as the middle of the range rather than its ceiling: the same bulletin reports a Virginia slate measured at 181 pounds per cubic foot, about three percent above the top of the band, which lifts every weight in the table with it.

Derived installed weight of the slate alone, at a 2.35:1 stone-to-coverage ratio (20 in slate, 3 in headlap), using the 159 to 176 lb per cubic foot band measured in the USGS Arkansas test series.
Nominal slate thickness	Stone volume per square	Derived weight per square	Derived weight per square foot	Where you meet this thickness
3/16 in	3.68 cu ft	585–645 lb	5.9–6.5 psf	The Park Service's description of historic standard-grade slate: units approximately 3/16 in thick, of consistent length and width.
1/4 in	4.90 cu ft	780–865 lb	7.8–8.6 psf	Reported in IIBEC Interface as the nominal thickness ASTM C406 has redefined as “standard”. The commonest specification for new work.
3/8 in	7.35 cu ft	1,170–1,295 lb	11.7–12.9 psf	Heavier textural work, and the middle courses of a graduated roof.
1/2 in	9.80 cu ft	1,560–1,725 lb	15.6–17.3 psf	The eave courses of a graduated roof, where the largest and thickest slates go. Thicker units than this exist.

Read this table one item at a time

3/16 in

Stone volume per square: 3.68 cu ft
Derived weight per square: 585–645 lb
Derived weight per square foot: 5.9–6.5 psf
Where you meet this thickness: The Park Service's description of historic standard-grade slate: units approximately 3/16 in thick, of consistent length and width.

1/4 in

Stone volume per square: 4.90 cu ft
Derived weight per square: 780–865 lb
Derived weight per square foot: 7.8–8.6 psf
Where you meet this thickness: Reported in IIBEC Interface as the nominal thickness ASTM C406 has redefined as “standard”. The commonest specification for new work.

3/8 in

Stone volume per square: 7.35 cu ft
Derived weight per square: 1,170–1,295 lb
Derived weight per square foot: 11.7–12.9 psf
Where you meet this thickness: Heavier textural work, and the middle courses of a graduated roof.

1/2 in

Stone volume per square: 9.80 cu ft
Derived weight per square: 1,560–1,725 lb
Derived weight per square foot: 15.6–17.3 psf
Where you meet this thickness: The eave courses of a graduated roof, where the largest and thickest slates go. Thicker units than this exist.

This is a derivation from published inputs, not a measurement and not a product figure. It assumes the geometry held constant at a 20 in slate and a 3 in headlap; a 4 in headlap on a shallower slope raises the ratio and the weight. The density band comes from one 1914 laboratory series on Arkansas slates; the same bulletin reports a Virginia slate at 181 lb per cubic foot, so a denser stone lands above every figure here. It counts the slate only — no underlayment, deck, fasteners, flashings, snow, or equipment. The governing number for a real project is the specific slate's published weight and a licensed engineer's assessment of the specific building.

What to do with that number

Compare it with the way the model rafter span tables are published: every one of them carries two dead-load cases, 10 psf and 20 psf. A ¼-inch slate covering at roughly 8 psf, plus deck, underlayment, fasteners, and flashings, is already pressing on the lower of those two assumptions before any snow lands on it. A graduated roof with ½-inch eave courses is not in the same conversation at all.

That is not a calculation you should finish yourself, and this page will not pretend otherwise. It is the calculation that tells you the conversation has to happen — with an engineer, about this building, in writing, before anyone orders stone.

Where it came from

Grade is a claim about a rock. Origin is a claim about a track record.Section link

Two slates can both be S-1 and behave nothing alike, because the standard tests properties and the field tests time. The Park Service's quarry figures are the closest thing to a time record that exists.

Slate is not manufactured to a recipe. It is metamorphic rock, and the Park Service explains the variability in one sentence: Slates vary in composition, structure, and durability because the degree to which their determinant minerals have been altered is neither uniform nor consistent. Local chemistry and the conditions of formation have produced a wide range of colors and qualities and ultimately determine the character of the slate found in these areas.

That is why quarry district carries information that a grade does not. The figures below are the Park Service’s, published in 1993 as a summary of field experience across the American slate districts. They are not test results and they are not warranties.

Service lives the National Park Service reports for U.S. slate districts in Preservation Brief 29, with the Brief's own caveat attached.
Slate district	Service life reported by the Park Service	Availability	What it means for a repair
Vermont and New York	About 125 years	Still quarried; the New York district around Granville holds one of the few commercial red slate deposits in the world.	The most straightforward matching problem in the country, and the source the Brief names as the closest available match for discontinued Pennsylvania ribbon slate.
Buckingham, Virginia	175 years or more	Still quarried.	Distinctive lustrous blue-black with high mica content. Difficult to imitate with anything else, which cuts both ways.
Pennsylvania Soft-Vein	In excess of 60 years	Historic; Pennsylvania was historically the largest producer of all.	Many roofs of this stone are now well past the reported range. A roof at year 100 in Soft-Vein is a different proposition from a roof at year 100 in Buckingham.
Pennsylvania Hard-Vein	Roughly 100 years	No longer quarried.	Matching means salvage. Plan repairs around the stock you can secure, not around a catalogue.
Peach Bottom	At least 200 years	No longer quarried. The district contained the first commercial slate quarry in the United States, opened in 1785.	The longest-lived American slate in the Brief's account, and unobtainable. A Peach Bottom roof is worth repairing carefully because it cannot be replaced in kind.

Read this table one item at a time

Vermont and New York

Service life reported by the Park Service: About 125 years
Availability: Still quarried; the New York district around Granville holds one of the few commercial red slate deposits in the world.
What it means for a repair: The most straightforward matching problem in the country, and the source the Brief names as the closest available match for discontinued Pennsylvania ribbon slate.

Buckingham, Virginia

Service life reported by the Park Service: 175 years or more
Availability: Still quarried.
What it means for a repair: Distinctive lustrous blue-black with high mica content. Difficult to imitate with anything else, which cuts both ways.

Pennsylvania Soft-Vein

Service life reported by the Park Service: In excess of 60 years
Availability: Historic; Pennsylvania was historically the largest producer of all.
What it means for a repair: Many roofs of this stone are now well past the reported range. A roof at year 100 in Soft-Vein is a different proposition from a roof at year 100 in Buckingham.

Pennsylvania Hard-Vein

Service life reported by the Park Service: Roughly 100 years
Availability: No longer quarried.
What it means for a repair: Matching means salvage. Plan repairs around the stock you can secure, not around a catalogue.

Peach Bottom

Service life reported by the Park Service: At least 200 years
Availability: No longer quarried. The district contained the first commercial slate quarry in the United States, opened in 1785.
What it means for a repair: The longest-lived American slate in the Brief's account, and unobtainable. A Peach Bottom roof is worth repairing carefully because it cannot be replaced in kind.

Preservation Brief 29 attaches its own caveat to these numbers: “The life spans provided should be used only as a general guide in determining whether or not an existing slate roof is nearing the end of its serviceable life.” The service lives and the availability column both describe the position as the Brief recorded it in 1993, and they describe stone, not roofs — a Vermont slate roof with corroding nails will not reach 125 years, and a well-maintained one may pass it. Check current availability with a supplier before planning a repair around it.

Ribbons, and why some slate was retired

Ribbons are visible bands on the cleavage face marking geological periods when more carbonaceous matter, calcite, or coarse quartz was present in the sediment. They weather faster than the clear stone, they were commonest in Pennsylvania quarries, and ribbon slate is no longer manufactured for roofing purposes. ASTM C406 makes the same judgement in specification language, excluding slates containing soft carbonaceous ribbons because their service life is uncertain under some conditions of use. If your roof has ribbons in it, you are looking at stone the industry has already retired.

Imported slate is a genuinely open question

Slate from China, Spain, Africa, and elsewhere has been imported into the U.S. market for decades. The Park Service, writing in 1993, advised limiting imported slate to new construction because its colours and textures often do not match U.S. stone. The more current concern is documented in IIBEC Interface: ASTM C406 contains no test for oxidation, so a slate can meet the standard and still develop rust staining after installation from iron pyrite inclusions, and the same article reports warped slates and slates with knots and cramps reaching the market.

This is not an argument that imported slate is bad. It is an argument that the certificate does not cover the failure mode, and that on a covering whose entire value proposition is the century you are not going to spend replacing it, an absent multi-decade field record in your climate is a real and unpriced risk. Ask what the quarry is, ask what installations of that quarry’s stone exist in your region, and ask how old they are.

From the ground

How to tell a fastener problem from a stone problem without leaving the groundSection link

This is the distinction that decides whether you are buying a repair campaign or a replacement, and it can be made from a driveway with a pair of binoculars and a camera.

The Park Service publishes a nine-point repair-or-replace guideline in Preservation Brief 29. Several of its points can be worked from outside the building, and they separate the two failure families cleanly. Where the Brief’s own points would put someone on the roof or in the attic, they are either left out here or reframed for a slate that has already come down.

Signs the fasteners have failed and the stone has not

Slates sliding out of position — dropped below their course line, or gone entirely — scattered across a slope rather than grouped. The Brief’s own reading of this: it may be that ferrous metal fasteners were used and that these are corroding, while the slates are still in good condition. Its prescription is to salvage the slates and re-lay them.
Slates on the ground that are whole and unbroken. A slate that failed as stone shatters or crumbles; a slate that lost its nails arrives intact.
The rest of the slope looking sound, with no general shedding or flaking visible in a zoomed photograph.

Signs the stone itself is at the end

Visible flaking and scaling — the Brief describes paper-thin laminations coming off the surface as slate delaminates.
Slates that feel brittle and will crack under hand pressure, or that give a dull thud rather than a clear ring when tapped. Both tests are performed on a slate that has already come off the roof and is on the ground — never on the roof, and never by you at height.
Deterioration concentrated at eaves and valleys first, which is where concentrated flow and ice damming do their work.

The 20 percent line

The Brief gives a threshold rather than a judgement call: if 20% or more of the slates on a roof or roof slope are broken, cracked, missing, or sliding out of position, it is usually less expensive to replace the roof than to execute individual repairs, and it notes that on an older roof even an experienced slater will break additional slates while attempting them. Historic Environment Scotland arrives at the same place from a different tradition, putting the reslate threshold at 20 to 25 percent of the covering affected.

Two consequences worth holding on to. First, the threshold is applied per slope, not per roof — the Brief notes that deterioration is often not uniform, that one slope may need replacement while others are repaired, and that this is how the cost of replacement gets spread over years. Second, below the threshold, prompt repair is not just cheaper than replacement, it is what makes the long service life real.

Three things to photograph before you call anyone

The valleys and the chimney, zoomed. On a slate roof these are where the leak usually is, and they are metal.
Any black smears of mastic. They are a record of past repairs done wrong and a strong hint about what is underneath.
The slope with the worst exposure — often, though not always, the southern one, which the Brief associates with more rapid deterioration. Comparing slopes tells you whether you have a whole-roof problem or a one-slope problem.

Then use the roof inspection page to work out what a real inspection should cover, and the triage hub to sequence the decision. If water is actively coming in, start at emergency roof leaks and tarping instead — urgency first, economics second.

Cost and lifecycle

This page does not publish a dollar figure. Here is what it publishes instead.Section link

Slate labour markets are thin, regional, and unlike the asphalt market in kind rather than degree. There is no defensible national installed-cost dataset for slate, so instead of inventing one, this section gives the labour input that drives it and the two structural facts that decide the rest.

Clear roof expanse: 2–3 squares per dayAn experienced slater and one helper, on plain surface with no interruptions.
Complex roof: below 1 square per dayWhere chimneys, dormers, and valleys are in the way. Every interruption is hand-cut stone and hand-formed metal.
Planning average for a whole job: about 1 square per dayThe Park Service's own figure for estimating job duration, inclusive of flashings, gutters, and scaffolding set-up and break-down.

Units: Roofing squares (100 sq ft of covered roof) laid per crew-day by an experienced slater and one helper
Scope included: Laying slate, installing flashings and gutters, and setting up and breaking down scaffolding
Not included: Tear-off of the existing roof, structural evaluation or reinforcement, deck replacement, permits, disposal, and all materials
Geography: United States
Data as of: Spring 1993 — the publication date of NPS Preservation Brief 29
Confidence: Low to moderate as a present-day number. It is a labour-productivity figure from a 1993 government technical publication, not a market rate, and it says nothing about what a crew-day costs in any market today. It is published here because it is sourced and because it explains the shape of a slate price, not because it can be multiplied into a quote.
Method: How this figure is built

Read the three figures above together and the economics of slate stop being mysterious. A roof that a shingle crew covers in a day takes a slate crew a week, the work cannot be accelerated with a nail gun, and the people who can do it are scarce. That is the price. It is labour, not stone.

The comparison that surprises people

The Park Service makes a point about look-alike products that runs against the usual intuition. Artificial slate costs less than natural slate as a material, but the total initial cost of an artificial slate roof is only marginally less than a natural slate roof, because all the other costs associated with replacing a slate roof, such as the cost of labor, flashings, and tearing-off the old roof, are equal in both cases. The material is the small half of the invoice.

That observation was written in 1993 about the mineral-fibre products of that era, and it should not be read as a verdict on the polymer composites sold today, which are lighter, are made differently, and are covered on our composite roofing page. The transferable part is the structure of the cost, not the conclusion about a product category that has since changed.

Two things that move a slate price more than the slate does

Whether the structure needs work. A structural evaluation is a professional fee; structural reinforcement is a construction project. Neither belongs in a roofing line item, and on a conversion both may be unavoidable.
Whether the roof can be recovered. Usually it cannot. Widely adopted model reroofing text prohibits a roof recover where the existing covering is slate, clay, cement, or asbestos-cement tile, so wherever that text has been adopted a slate roof is a tear-off. Whether it has been adopted where you live, in which edition, and with what amendments is a question for your authority having jurisdiction rather than for this page. The Park Service gives the structural reason as well as the code one: the old slate should come off to prevent overloading of the roof timbers.

When you are ready to attach numbers to any of this, the cost hub explains what a range means before a figure is put on it, and the calculators show their formulas rather than their conclusions.

A planning range is not a quote. It is a number to argue with a proposal about — 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

This is the first question, not the fifth. The derivation above puts a slate covering somewhere around 6 to 17 pounds per square foot depending on thickness — before underlayment, deck, fasteners, flashings, or snow. Every model rafter span table in IRC Chapter 8 is published with two dead load cases — 10 psf and 20 psf — which is the plainest available statement that the framing under a house was sized against an assumption about what the roof would weigh.

Two situations, two different conversations. Re-slating a roof that has carried slate for eighty years is a low-risk project where the building has already demonstrated the answer. Converting an asphalt roof to slate is a structural project. Even on a re-slate, the Park Service directs that the timbers be checked for deflection, cracking, and twisting, and that if such conditions are found, a structural engineer experienced in working with older buildings should be consulted.

Jurisdictions write that trigger into their own permit rules in their own ways. The City of Santa Clarita, California, for one, may require a licensed engineer’s or architect’s structural calculations where the new roof covering exceeds 7 pounds per square foot — a line the ordinary ¼-inch slate in the derivation above has already crossed, before any deck or underlayment is counted. Whether your jurisdiction sets a threshold, and where, is a question for it and not for this page.

A change to slate is a structural question and it is answered by a licensed engineer looking at this building — the species, grade, size, spacing, span, and condition of the actual framing — not by any weight figure on this or any other page.

Slope and drainage

The model code confines slate to 4:12 and steeper, and increases headlap as the slope falls. Trade installation guidance adds that headlaps may be increased where ice damming or poor drainage is expected. If part of your roof is below 4:12 — a porch, a connector, a dormer top — that part is a different assembly with a different material, and it should appear as a separate line in a proposal rather than being absorbed into “the slate roof”. Our roof pitch page explains how to establish slope without going up.

Code and jurisdiction

There is no nationwide building code for site-built houses in the United States. Everything quoted here from the 2024 IRC is model text that a jurisdiction may adopt, amend, delay, or decline, and re-roof permits are inspected differently from new construction in many places.

The provisions that most often matter on a slate job are R905.6.2 (4:12 minimum slope), Table R905.6.6 (headlap by slope, and two fasteners per slate), R905.6.4 (compliance with ASTM C406), R905.6.7 (flashing, including a valley flashing not less than 15 inches wide), and the reroofing text in Section R908 that prohibits recovering over an existing slate roof. That last provision is numbered R908.3.1.1 in the 2021 IRC and is renumbered between editions and between the residential and commercial codes, so cite it by its words rather than by a number.

The International Code Council publishes the authoritative version of that model text; the source list below links it directly. A commercial code aggregator is a convenient way to read the same words, but it is not your jurisdiction’s adopted law and neither is the model text itself.

What an adoption actually looks like: the City of Santa Clarita, California, requires reroofing work to comply with chapter 9 of the 2019 California Residential Code, chapter 15 of the 2019 California Building Code, and the city’s own amendments — and its list of coverings that must be stripped rather than recovered reads wood shake, wood shingle, slate, clay, cement, or asbestos-cement tile, which is longer than the model list. One city, one edition, one date, and already not the model text. Yours will differ again.

Record your jurisdiction, its adopted edition, its amendments, and the effective date, and confirm with the authority having jurisdiction before relying on any dimension here. A citation to model IRC text is a citation to a model provision, not to the law where you live.

Fire

Slate is stone and does not burn, and that fact tempts almost everyone into the wrong conclusion. A Class A, B, or C fire classification belongs to a tested assembly — deck, underlayment, and covering together under a defined test — not to a material in isolation, and the history of slate in the model codes proves the point better than an argument could.

As documented in IIBEC Interface, slate roofs were classified Class A regardless of substrate before 2009; the 2009 and 2012 IBC editions accepted slate as a Class A assembly only over non-combustible decks such as concrete; and after fire testing in 2010 the 2015 IBC reinstated the Class A classification for slate over combustible decks with an approved underlayment. The stone did not change in those years. The tested assembly and the code edition did.

Fire classification applies to a tested assembly, not to a covering by itself. If a classification matters to you — for a wildland-urban interface property, an insurer, or a jurisdiction that requires one — the question is which tested assembly is specified and installed, and under which adopted code edition.

Wind

The 2024 IRC requires slate shingles to be tested in accordance with ASTM D3161, with the packaging labelled to show the classification, and the required classification set by a table in the code. That is a product-level laboratory classification and a labelling requirement. It is not a wind-design determination for your building.

Wind performance is site- and building-specific. Basic wind speed, exposure category, building height and geometry, pressure zone, enclosure classification, risk category, the attachment used, and the tested assembly all matter. A classification on a package — or an mph number in a brochure — is not a code determination for your roof.

Hail and impact

Impact classifications come from UL 2218, a laboratory steel-ball drop test. IBHS, which runs UL 2218 tests in its own laboratory on a device built to the standard’s specifications, describes the method precisely: the Class 4 projectile is a 2.00-inch steel ball, dropped from the height needed to match the kinetic energy a hailstone of that size would carry, striking each location twice. A failure is any evidence of opening — tearing, fracturing, cracking, or rupturing — on the back of the specimen. Crushed granules, dents, and openings visible only on the top face are not failures under the standard.

IBHS also states the limit of the whole exercise: these standardised tests do not exactly replicate natural hailstone impacts, and the pass criteria do not necessarily reflect the performance criteria used by insurance companies. For slate specifically, ask whether the product you are being offered has actually been classified under UL 2218 and at what class. Do not assume that stone is automatically in a class, and note that the Park Service records weathered slates as more prone to breakage, loss of corners, and cracking — an old slate roof and a new one do not behave the same under impact.

“Class 4 impact resistant” does not mean hail proof. It means one specimen of one product, in new condition, survived two strikes of a 2-inch steel ball without opening on the back face in a laboratory.

Climate

Slate resists frost because it is dense and absorbs very little water — the Park Service attributes its suitability for roofing directly to low porosity and low absorption. That protection is not permanent. Weathering increases absorption and reduces strength, so a slate that shrugged off freeze-thaw at year 20 may not at year 110, and the parts of the roof that weather fastest are the parts already carrying the most water.

The Park Service names them: areas of a roof subject to concentrated water flows and ice damming, such as along eaves and valleys, also tend to deteriorate more rapidly than other areas of the roof. On a slate roof in a cold climate, the eave and the valley are where the roof’s clock actually runs, which is why ice damming is a material question here and not only a comfort one.

Whether an ice barrier is required at your address, and how far up the roof it must run, depends on your jurisdiction's adopted code edition, its amendments, its climate designation, and the specific assembly. There is no universal rule and no number on this page is one.

Moisture and ventilation

A weathering slate roof can rot the deck underneath it without ever leaking. The Park Service describes the mechanism directly: the tendency of old, weathered slates to absorb and hold moisture can lead to rot in underlying areas of wood sheathing, and such rot can go undetected for long periods of time since, often, there is no accompanying leak. It also warns that condensation in a poorly ventilated attic produces false leaks that get blamed on the roof. Whether the assembly beneath is a vented attic or a correctly designed unvented one, both of which are legitimate, the diagnostic problem is the same — see ventilation.

Maintenance

Slate is low-maintenance, not no-maintenance, and the maintenance it needs is unusually consequential because deferring it is how a repairable roof becomes a replacement. The Park Service recommends inspecting overall condition annually and after severe storms, cleaning gutters at least twice in the fall and once in early spring, replacing damaged slates promptly, and having a professional experienced with slate inspect every five to seven years. It also recommends keeping a log book and filing bills and samples, which sounds fussy until a future owner needs to match a slate.

Access and site conditions

Everything a homeowner needs to know about a slate roof can be established from the ground, from a window, from photographs, or from a contractor’s documentation. The Park Service recommends, for safety reasons, that owners and maintenance personnel survey slate roofs from the ground with binoculars or from a cherry picker, and that foot traffic be kept off the roof entirely. Trade installation guidance says workers themselves are to avoid walking on slate and to stage the roof with planks, roof ladders, or hook ladders.

Do not climb onto a slate roof or into an attic to check anything on this page. Your weight breaks sound slates, the surface is slippery, and OSHA records falls as the leading cause of death for workers in residential construction — workers who are trained and equipped for it.

Asbestos, and the look-alike problem

Slate itself is not an asbestos product. But “artificial slate” and “mineral fibre slate” were sold for decades as substitutes, asbestos-cement shingles among them, and the reroofing code text that prohibits recovering over slate names slate, clay, cement or asbestos-cement tile in the same breath for a reason. Older felts, mastics, and roof cements found under and around a slate roof may also contain asbestos. EPA guidance is that the only way to be sure a material contains asbestos is laboratory testing, that testing is warranted where material is damaged or where a renovation would disturb it, and that samples should be taken by a properly trained and accredited asbestos professional.

If the building predates 1990 and you cannot positively identify the covering as natural stone, do not let anyone scrape, break, saw, or pressure-wash it. Testing before disturbance is the correct sequence, and state and local rules on notification and disposal vary.

Warranty and repair

There is usually no warranty doing the work you think it is doingSection link

Slate is quarried, not manufactured, and the documents that come with it behave differently from the documents that come with a shingle.

What a slate material warranty is usually about: A slate warranty is generally a statement about the stone: that it meets a grade under ASTM C406 and will not deteriorate within a stated period. That is a narrower promise than it sounds, because a slate roof rarely fails at the stone. Read what the document says about the remedy as carefully as the term — a replacement slate delivered to your driveway in year 60 is not the same thing as a roof.
Grade is not a warranty, and a warranty is not a grade: The S-1 / S-2 / S-3 classification is a laboratory result on sampled material against ASTM C406’s thresholds. The expected service lives associated with those grades are the trade’s convention, not contractual terms. A proposal that offers “S-1 slate” is telling you about a specification; a proposal that offers a warranty is telling you about a document. Ask for both, and ask for the quarry.
The workmanship half is the half that matters: Nail metal, nail depth, headlap, flashing metal and weight, and whether one metal was used throughout are all workmanship decisions, and they are what decides when the roof ends. Read the installer’s workmanship warranty for its length, what triggers a callback, whether a diagnostic visit is chargeable, whether it survives a sale, and what happens if the company stops trading. A ten-year workmanship warranty on a hundred-year roof is a statement about the company, not about the roof.
What a document cannot fix: Nothing in any warranty produces a person capable of repairing a slate roof in your area in year 40. Availability of skilled labour is the risk that no document on a slate project addresses, and it is the one most likely to determine the outcome.

Repairability

Slate is highly repairable in the hands of someone who has done it, and highly damageable in the hands of someone who has not. Both halves of that sentence are load-bearing.

A correct single-slate repair is a defined procedure. The old nails are cut or pulled with a slate ripper, a matching slate is slid in and secured with one nail through the vertical joint of the course above, and a copper bib roughly 3 inches by 8 inches is slid up under the joint to cover the new nail hole. Metal hooks, clips, and straps bent over the tail are alternatives. None of that requires exotic equipment; all of it requires knowing it.
Mastic is not a repair. The Park Service is unambiguous that roofing mastic and sealants are not a viable repair alternative, because they harden and crack, make future repairs harder, are unsightly, and when applied to metal flashings, accelerate their corrosion. Visible black smears on a slate roof are a record of past bad repairs and a fair reason to expect more of them underneath.
Matching is real work. Size, shape, colour, texture, exposure, and coursing all have to match, and some slates cannot be matched at all — ribbon slate is no longer manufactured, and whole quarry districts have closed. On a graduated or textural roof, one stock size does not patch anything.
Salvage is often the right answer. Where corroded nails or worn flashings caused the failure, the Park Service notes that salvaging at least some slates should be possible, each sounded first for cracks. On a partial replacement, salvaging is how you buy repair stock for the slopes you are keeping.
Weathered slates cannot be flipped. Delamination is as bad or worse on the underside, because that is where the gypsum concentrates. A contractor proposing to turn the old slates over has misunderstood the mechanism.
Repairs beget repairs. Even an experienced slater will break additional slates working on a brittle old roof, which is why the Park Service pairs its case-by-case guidance with a blunt numerical threshold.

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

None of these require you to know roofing. Every one of them is answerable in a sentence by someone planning to do the work properly, and evasively by someone who is not.

What nail are you using — what metal, what gauge, and what length?
The single highest-consequence question on a slate roof, and the one least likely to be in the proposal. “Copper” or “Type 304 stainless” is an answer. “Roofing nails” is not. Trade guidance sets nail length at roughly twice the slate thickness plus an inch, fully embedded in the deck.
What quarry is the slate from, what grade, and can I see the ASTM C406 documentation?
Quarry of origin carries more information about service life than colour does, and the Park Service’s own figures for U.S. districts range from “in excess of 60 years” to “at least 200”. A supplier who cannot name the quarry cannot tell you what you are buying.
What headlap are you laying, and how did you arrive at it?
The answer should reference the slope of the roof. Model code sets 4 inches below 8:12, 3 inches from 8:12 to 20:12, and 2 inches above that. If different slopes on the building get the same headlap, ask why.
What metal are the flashings, what weight, and is it the same metal throughout including the gutters?
Mixing metals designs galvanic corrosion into the roof on day one. The Park Service directs using a single metal for all flashings and the rainwater system, names 16 oz copper as the minimum weight, and calls for 20 or 24 oz in gutters, valleys, and areas with limited access.
Has an engineer looked at the framing, and what did they say in writing?
On a conversion to slate this is not optional. On a re-slate it is still worth asking, because the Park Service directs checking the timbers for deflection, cracking, and twisting during the stripping.
How many slate roofs have you laid in the last five years, and can I see one and speak to its owner?
Trade guidance recommends that foremen or supervisors have at minimum five years of professional experience installing slate. A slate roof laid by a crew learning on your building will look acceptable for several years and then stop being acceptable.
How will you stage the roof so nobody walks on the slate?
Planks, roof brackets, hook ladders, or chicken ladders is the right answer. Walking on the finished slate is how a new roof acquires cracked units that will not announce themselves for years.
Are you nailing this by hand?
It should be yes. Trade guidance for slate does not permit pneumatic or electric nail guns, because a gun cannot leave a nail set to hang the slate rather than clamp it.
If a slate breaks in ten years, who repairs it, and are you leaving me spare slate?
A slate roof needs occasional competent repair far more than it needs replacement. Leftover slate from this job is the only guaranteed colour and texture match you will ever have.
If the deck under the old roof is rotten, how is that priced, and what are you replacing it with?
The Park Service specifically warns against plywood as a replacement for deteriorated board sheathing under slate, because driving a nail into it bounces and loosens adjacent slates, and against pressure-treated lumber because it shrinks. The allowance and the unit rate belong in writing before work starts.

Require these in writing

Quarry of origin, grade against ASTM C406, colour, size, and nominal thickness of the slate.
Nail metal, gauge, and length, stated as a product rather than a category.
Headlap in inches, stated per roof slope where the building has more than one.
Flashing metal and weight for every location — valleys, chimneys, sidewalls, headwalls, dormers, and built-in gutters — and confirmation that one metal is used throughout including the rainwater system.
Underlayment specified by product, and the ice-barrier extent if one is required in your jurisdiction.
How the roof will be staged, and the explicit commitment that finished slate will not be walked on.
A deck-repair allowance with a unit rate, the replacement material named, and a method for documenting what was replaced.
Quantity of spare slate left on site at completion, and where it will be stored.
Whether any structural evaluation has been done, by whom, and what it concluded.
Pre-cover photographs of every flashing detail, delivered to you — this is the only inspection of the concealed work you will ever be able to make.

What goes wrong

Misconceptions and failure modesSection link

Common misconceptions

Common belief
A slate roof lasts a hundred years, so I will never touch it.
What is actually true
The slate may. The roof is a system, and the fasteners, flashings, gutters, and deck are on shorter clocks. A hundred-year slate roof is a roof that received competent repairs and at least one flashing campaign along the way. The Park Service recommends annual condition inspection and a professional inspection every five to seven years.
Common belief
Slates are sliding off, so the slate has failed.
What is actually true
Almost the opposite. The Park Service’s repair-or-replace guidance asks the question directly: are many slates sliding out of position? If so, it may be that ferrous metal fasteners were used and that these are corroding, while the slates are still in good condition. The prescribed response is to salvage the slates and re-lay them. Sliding is evidence about the nails, not about the stone.
Common belief
The roof is leaking, so the slate is leaking.
What is actually true
The Park Service’s guidance says plainly: do not assume the slates are leaking. Gutters, valleys and flashings are more likely candidates, and it adds that false leaks can come from attic condensation caused by poor ventilation. Even a pinhole in a flashing can permit large quantities of water to enter the building.
Common belief
Colour tells you what quality of slate you are getting.
What is actually true
It does not. Color permanence generally provides no indication of the durability of slate. Fading and unfading describe colour stability; they say nothing about how long the stone lasts. Quarry district and grade carry that information, and colour is what the marketing leads with.
Common belief
Slate is Class A, so it is a fire-rated roof.
What is actually true
Fire classification belongs to a tested assembly, not to a covering. Slate’s own history in the model codes demonstrates it: Class A regardless of substrate before 2009, Class A only over non-combustible decks in the 2009 and 2012 IBC, and Class A over combustible decks with an approved underlayment restored in the 2015 IBC after testing. Same stone, three different answers, three different code editions.
Common belief
We can just go over the slate to save the tear-off.
What is actually true
Widely adopted model reroofing text prohibits a roof recover where the existing covering is slate, clay, cement, or asbestos-cement tile. The Park Service gives the structural reason as well: the old slate should be removed to prevent overloading the roof timbers. Anyone offering to roof over your slate is offering something your jurisdiction has probably prohibited.
Common belief
Any competent roofer can repair slate.
What is actually true
Slate has its own tools — a ripper to cut concealed nails, a slate hammer, a cutter, a stake — and its own rules about nail depth, bibs, and pyramids. The Park Service states that the installation and repair of slate roofs should be entrusted only to experienced slaters. The most common damage to old slate roofs is done by well-meaning trades walking on them and sealing them.

How it actually fails

Nail sickness — fastener corrosion under sound slate: Ferrous or galvanized nails corrode, the nail holes wear, and slates begin to slip out one after another while the stone remains good. Historic Environment Scotland: as more and more nails decay, numerous slates may become loose and slip out of place. The Park Service records the rusting of nineteenth-century cut nails as a common cause of slate loss.What you can see: Slates visibly out of line, dropped an inch or two below their course, or missing entirely — usually scattered across a slope rather than concentrated in one spot. Fragments of intact, unbroken slate on the ground. The pattern accelerates: a few slates a year becomes a few a month.
Flashing failure under a roof with decades left in it: Valleys, chimney sets, and built-in gutters are metal, and metal is the weakest point in any roof. Light-gauge or mixed metals corrode, pinhole, and open at seams long before the slate is finished, and replacing valley metal means taking slates off both slopes.What you can see: Leaks that track heavy or prolonged rain rather than every shower, staining running as a line below a valley, and damp around a chimney chase. From the ground, green streaking or corrosion product on masonry below metal.
Delamination — the stone genuinely weathering out: Calcite and iron sulphide impurities react with wet-dry and hot-cold cycling to form gypsum, which occupies about twice the volume of the calcite it replaced. The internal stress splits the slate along its cleavage planes. It is worse on the underside than on the face, because that is where the gypsum concentrates.What you can see: Paper-thin flakes shedding from the surface. Slates that feel soft or spongy and crack under hand pressure. A dull thud rather than a clear ring when a loose slate is tapped — the Park Service’s own field test, performed on a slate that has already come off, never on the roof.
Nails driven too tight, or left projecting: A slate is meant to hang on the nail shank, not be clamped by the head. Over-driven nails crack the slate on installation or in the first thermal cycles; under-driven nails leave a proud head that punctures the slate lying over it. A nail gun does this reliably, which is why trade guidance for slate does not permit one.What you can see: Cracked slates appearing in the first few years of a new roof, often in a line rather than at random. Neat round punctures in individual slates, visible in a zoomed photograph.
Foot traffic: Chimney sweeps, painters, antenna installers, gutter cleaners, and roofers working on something else walk the slate. The Park Service lists the weight of a workman walking on the roof among the ordinary causes of individual slate damage, and it recommends keeping foot traffic off entirely.What you can see: Broken slates clustered along an obvious route — from a ladder point toward a chimney or a vent. Damage that appears after other trades have visited rather than after weather.
Mastic as maintenance: Cracks and slipped slates are smeared with roof cement instead of being repaired. It works briefly, then hardens and cracks, and on metal flashings it accelerates corrosion — so the repair actively shortens the life of the component next to it.What you can see: Black smears visible on the roof from the ground or in photographs. A repair history of recurring leaks in the same place. Expect to find more of it underneath.
Deck rot with no leak: Old, weathered slates absorb and hold moisture, and the wood under them rots from persistent damp rather than from running water. The Park Service notes such rot can go undetected for long periods of time since, often, there is no accompanying leak.What you can see: Nothing at all from inside, usually. It is found when slates start letting go because the deck no longer holds a nail, or when someone finally opens the roof up.
Insufficient headlap or badly broken joints: Where slates lap the joints in the course below by less than the required amount, water passes between the joints and through the nail holes to the felt. Insufficient headlap lets wind-blown water travel past the heads of the slates in the course below.What you can see: Leaks that appear only in driven rain, spread over an area rather than at one point, on a roof that otherwise looks sound. Almost always a construction defect rather than an ageing one, and it shows up early.

Sources and further readingSection link

Understanding Roofing / Published August 14, 2026 / Updated August 26, 2026

Scope and limitations

It cannot tell you whether your building can carry a slate roof.
That is a structural determination made by a licensed engineer who has looked at the actual framing — species, grade, size, spacing, span, condition, and what else the structure is already carrying.
It does not publish an installed cost.
No transparent national dataset for slate installation was found, and slate labour markets are thin and regional enough that a national dollar range would be an invention rather than an estimate.
The labour-productivity figures published instead come from a 1993 government technical publication and are not a market rate.
It cannot tell you how long your slate will last.
The service-life figures here are planning ranges attached to quarry districts and to laboratory grades, and the Park Service says of its own numbers that they should be used only as a general guide.
It cannot tell you what your jurisdiction requires.
Every code dimension here is model text from the 2024 IRC, which a government may adopt, amend, or decline.
Your adopted edition, its amendments, and your authority having jurisdiction govern.
It does not reproduce the numeric thresholds in ASTM C406 for absorption and depth of softening.
The standard is not published free of charge, and no free source for those values could be verified at the correct hierarchy level.
The grade names, the three underlying test methods, and the service lives the trade associates with each grade are sourced; the thresholds are not, and are therefore not stated.
Its weight figures are a derivation, not a measurement.
They are calculated from published stone densities and code headlap, and they are shown with their inputs so the arithmetic can be checked.
The governing figure for a real project is the specific slate's published weight and an engineer's assessment of the building.

Preservation Brief 29: The Repair, Replacement, and Maintenance of Historic Slate Roofs
Jeffrey S. Levine, U.S. National Park Service, Technical Preservation Services / Spring 1993
The 60-to-125-year service-life range and the over-200-year outliers; quarry-district service lives for Vermont/New York, Buckingham Virginia, Pennsylvania Soft-Vein and Hard-Vein, and Peach Bottom; that colour permanence gives no indication of durability; the delamination mechanism through calcite-to-gypsum conversion and why slates cannot be flipped; that non-ferrous nails must be used because plain steel and galvanized rust out long before the slate does, and that rusting cut nails are a common cause of slate loss; that slate nails must not be driven tight; the exposure formula and the 3 in standard headlap with its slope variations and the 4:12 floor; that flashings are the weakest point in any roof and often wear out before the slate; single-metal flashing to avoid galvanic action and the 16/20/24 oz copper weights; the repair sequence with ripper, bib, and pyramids; that mastic is not a viable repair and accelerates flashing corrosion; the 20 percent repair-or-replace threshold and the nine-point repair/replace guideline including sliding slates as evidence of corroding ferrous fasteners; deck rot without a leak; false leaks from attic condensation; ground-and-binoculars survey and keeping foot traffic off; the 2-3 squares, under-1 square, and about-1-square-per-day labour figures; that old slate must come off to prevent overloading the roof timbers and that a structural engineer should be consulted where stress is found; the plywood and pressure-treated sheathing warnings; and the artificial-slate total-cost observation.
Written for historic buildings in 1993. It is technical guidance, not a code determination anywhere, and its quarry service lives are its author's summary of field experience at that date, offered explicitly as a general guide. It predates the current ASTM C406 revisions and today's imported-slate market.
2024 International Residential Code, Chapter 9: Roof Assemblies — R905.6 Slate shingles (R905.6.1, R905.6.2, R905.6.4, R905.6.5, Table R905.6.6, R905.6.7) and Section R908 Reroofing
International Code Council, ICC Digital Codes — the publisher's own edition of the model text / 2024 edition
That slate shingles are fastened to wood structural panels or solid lumber sheathing; the 4:12 minimum slope; that slate shingles shall comply with ASTM C406; that slate shingles shall be tested in accordance with ASTM D3161 and labelled with the classification; the headlap-by-slope table of 4 in, 3 in and 2 in; two fasteners per slate; the 15 in minimum valley flashing width; and the reroofing provision that a roof recover shall not be permitted where the existing covering is slate, clay, cement or asbestos-cement tile, where it is water-soaked or deteriorated, or where two or more applications already exist.
MODEL text, not adopted law. It is not the law in any jurisdiction until that jurisdiction adopts it, and adoptions routinely amend, delay, or decline provisions — which is why no jurisdiction or effective date is claimed here. The section text quoted on this page was read in full through the free cross-jurisdiction reproduction listed below and checked against this publisher's own edition. Confirm the adopted edition, its amendments, and its effective date with your authority having jurisdiction.
2024 International Residential Code, Chapter 8: Roof-Ceiling Construction — R802.4.1 and Tables R802.4.1(1) through R802.4.1(8)
International Code Council, ICC Digital Codes — the publisher's own edition of the model text / 2024 edition
That rafters are sized from published span tables, and that each of those tables carries two dead-load cases, 10 psf and 20 psf — the basis for the statement that existing framing was sized against an assumption about roof weight.
MODEL text, not adopted law, and prescriptive tables at that. They do not substitute for engineering analysis of an existing building, which is what a change to a heavier covering requires. The table titles and their dead-load cases were read through the free cross-jurisdiction reproduction listed below.
Roof Recovering and Replacement — Residential (Building & Safety public information sheet, revised 1 January 2020)
City of Santa Clarita, California — Building & Safety Division / Revised 1 January 2020
One named adoption, with an edition and a date: this jurisdiction requires reroofing work to comply with chapter 9 of the 2019 California Residential Code and chapter 15 of the 2019 California Building Code plus the city's own amendments; it requires existing coverings to be removed where the existing covering is “wood shake, wood shingle, slate, clay, cement, or asbestos-cement tile”, citing CBC section 1511.3.1 — a list amended longer than the model text; and it may require a licensed engineer's or architect's structural calculations where the new roof covering exceeds 7 pounds per square foot.
The requirements of one California city, on one date, under one code cycle. It is cited here as evidence that adoptions exist, differ, and amend the model text — not as a requirement anywhere else, and not as current. It is a permit-counter information sheet rather than the adopted ordinance itself; check the city's own current sheet and code cycle before relying on it, and your own jurisdiction for what governs you.
Cross-jurisdiction reproductions of the IRC roof-assembly, roof-ceiling, and reroofing provisions
UpCodes — a commercial code aggregator, not a jurisdiction or a code publisher
Nothing on its own. It is the free reading copy through which the ICC provisions cited above were read in full, and it is recorded here so the reader can open the same words without a subscription. It also shows how far the section numbering for the recover prohibition moves between editions and between the residential and commercial codes.
A commercial republication. It is neither adopted law nor the code publisher's own edition, the jurisdiction and section number it surfaces will not always be the ones that govern where you live, and what it surfaces can change. No claim on this page rests on it alone.
ASTM C406/C406M-22, Standard Specification for Roofing Slate
ASTM International / 2022 revision
That the specification covers the material characteristics, physical requirements, and sampling appropriate to selecting slate for use as roof shingles; that roofing slate is classified as Grade S1, S2 or S3, each meeting specified breaking load, absorption, and depth-of-softening requirements; and that slates containing soft carbonaceous ribbons are excluded because their service life is uncertain under some conditions of use.
The publicly visible page is the scope and abstract only. The numeric thresholds for each grade are inside the paid standard and are therefore not quoted on this page. The standard describes tests on sampled stone; it makes no claim about an installed roof.
Slate Roofing: Innovation in an Old Industry, Part II — Standards, Codes, Testing, and More
Julie Palmer, IIBEC Interface (International Institute of Building Enclosure Consultants), February 2017 / February 2017
The expected service lives the trade associates with each ASTM C406 grade — over 75 years for S-1, 40 to 75 for S-2, 20 to 40 for S-3; that the underlying tests are ASTM C120 flexure, C121 water absorption and C217 weather resistance; the 2005 change to a minimum 575 lbf breaking load in place of the previous 9,000 psi modulus of rupture; that industry guidance is to use only S-1 for roofing; that C406 contains no oxidation test, so compliant slate can still develop rust staining; that a revision of C406 redefined nominal 1/4 in as the “standard” thickness; that warped slates and slates with knots and cramps have reached the market; and the IBC fire-classification history for slate across the 2009, 2012 and 2015 editions.
Qualified trade and consulting guidance, not adopted code and not the standard itself. Its account of code changes is a summary; the adopted text in your jurisdiction governs. Written in 2017, before any later revision of C406 or the model codes.
General Installation Guidelines for Natural Quarried Roofing Slate, version 1.0.5
Slate Roofing Contractors Association of North America — adopted by unanimous vote of the SRCA Board of Directors, 4 April 2012; the copy read here is hosted on the website of the Vermont Slate Company / 4 April 2012
Nail specification — solid copper smooth-shank roofing nails at minimum 11 gauge and 1.25 in, or Type 304 stainless, with length approximately twice the slate thickness plus one inch, fully embedded in the deck; that nails are driven so the head lies within the countersunk hole and not so far as to strain the slate; that pneumatic and electric nail guns are not permitted; two nails per slate placed above the head of the underlying slate; headlap by slope with 4:12 as the recommended floor and increased headlap in ice-dam-prone areas; that slates are not to be bedded in mastic; minimum 16 oz copper flashing with 20 oz recommended; that workers are to avoid walking on slate and are to stage the roof; the five-years-of-experience recommendation for foremen; and the published exposure and slates-per-square table that the worked example on this page is checked against, including 8½ in exposure and 170 slates per square for a 10 by 20 in slate at 3 in headlap.
Trade installation guidance hosted by a slate supplier. Its own disclaimer states that it does not supersede local building codes and that the publisher does not warrant its accuracy for specific projects. It is not adopted code and it is not a substitute for a specific product's instructions.
Slate in the United States, Bulletin 586
T. Nelson Dale and others, U.S. Geological Survey, 1914 / 1914
Measured densities of U.S. roofing slate. Its laboratory series on Arkansas slates reports weights per cubic foot of about 159 to 176 pounds, at apparent specific gravities of about 2.54 to 2.82 and true specific gravities of about 2.70 to 2.87; separately it reports a Virginia slate at specific gravity 2.90 and 181 pounds per cubic foot. The Arkansas band is the density input to the weight derivation on this page and the Virginia figure is why that band is presented as a middle rather than a ceiling.
A 1914 geological survey. Its density measurements remain sound because the rock has not changed, but its industry, market, and quarry information is historical. The only weight-per-cubic-foot test series it publishes is the Arkansas one, so the band on this page is one region's stone rather than a national average. It contains no weight-per-square table for finished roofs, which is why the figure on this page is derived rather than quoted.
Relative Impact Resistance of Asphalt Shingles: Summary of UL 2218 Impact Tests
Insurance Institute for Business & Home Safety / August 2014
The UL 2218 method — Class 1 through Class 4 steel balls of 1.25, 1.50, 1.75 and 2.00 in diameter, dropped from the height needed to match the kinetic energy of a hailstone of the same size, with two strikes at each location; that a failure is any opening on the back of the specimen and that crushed granules, dents, and top-surface-only openings are not failures; that the standardised tests do not exactly replicate natural hailstone impacts; and that the pass criteria do not necessarily reflect the criteria insurers use after a hail event.
The test programme reported here is on asphalt shingles, not slate. It is cited for the method and its stated limits, not for any performance figure for slate. Whether a particular slate has been classified under UL 2218 is a product-specific question for its supplier.
Slate Roofs
The Engine Shed, Historic Environment Scotland
The definition of nail sickness — that it occurs when the nails rust through and causes individual slates to slip, and that as more nails decay numerous slates become loose and slip out of place; the guidance that where more than 20 to 25 percent of the covering is affected it usually makes more sense to reslate the whole roof than to replace individual slates; and that scarcity of matching slate becomes a constraint on repair once a quarry district closes.
Scottish heritage guidance, written about Scottish slate, Scottish construction traditions, and Scottish practice. It is cited here for the failure mechanism and the term, both of which transfer. Its thresholds, materials, and regulatory references do not apply in the United States.
How do I know if I have asbestos in my home?
U.S. Environmental Protection Agency
That the only way to be sure whether a material contains asbestos is to have it tested by a qualified laboratory; that EPA recommends testing suspect materials where they are damaged or where a renovation 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.
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. That trained workers use fall protection is a reason for an untrained reader to stay off the roof entirely, not a procedure to copy.

Suggest a correction