For homeowners, building owners, and buyers in hot and high-sun regions

Heat and UV never break a roof. They age it, every day, for its whole life.

Steep-slope and low-slope · single-family, multifamily, and light commercial

There is no event to point at and no claim to file. The damage is the sum of ten thousand hot afternoons, and it is the reason the same covering reaches the bottom of its planning range in Phoenix and the top of it in Seattle.

30-second answer

What does extreme heat and ultraviolet light actually do to a roof, and is a cool roof worth it?

Heat and UV age a roof continuously rather than damaging it in one event. Sunlight drives the surface far above air temperature, ultraviolet light oxidises the exposed asphalt or polymer, and the daily heat-up and cool-down works fasteners, seams, and laps. A reflective roof lowers surface temperature and cooling bills, but how much it is worth depends on climate zone, insulation, and where the ducts run.

Safety first

A sunlit roof is a burn surface and a heat-illness risk before it is anything else.

The Department of Energy’s Building America Solution Center records that standard or dark roofs can reach 150 °F or higher in hot climates. That is a contact-burn surface. It is also an environment that produces heat illness quickly in anyone standing on it, and roofing work in that environment is why professional crews change their hours in summer rather than their pace.

The roof is not the only hazard in a heat event, and it is not the most urgent one. During an excessive-heat warning the people in the building matter more than the covering over them. Cooling-load arguments on this page are a planning conversation for a mild week, not something to act on during one.

OSHA lists the signs of heat stroke as confusion, slurred speech, unconsciousness, seizures, very high body temperature, a rapid heart rate, and skin that is either heavily sweating or hot and dry — someone still sweating hard can be having one. If you see those signs in anyone working on or near a roof, call 911 immediately, move them to shade or air conditioning, cool them with cold water or ice while you wait, and do not leave them alone. Do not send anyone onto a roof to check what this page describes — every observation here can be made from the ground, from a window, from a photograph, or from a report.

Learning paths and saved lessons

At a glance

The short versionSection link

Every figure below is a measured or published value with its source named in the source list at the foot of this page. None of them is a determination about a specific building, and none of them is a code requirement anywhere — see the code note for why cool-roof rules are a jurisdiction-by-jurisdiction question.

The mechanism: Continuous ageing, not a single failure eventNothing here produces a date of loss. Heat and UV move a covering down its planning range gradually, which is exactly why the damage is easy to sell against and hard to claim for.
Surface against air: 50 to 90 °F hotter than the airEPA: on a hot, sunny summer day the sun can heat dry, exposed surfaces like roofs and pavement to temperatures 50 to 90 °F hotter than the air. DOE's BASC puts dark roofs at 150 °F or higher in hot climates.
What UV attacks: The exposed asphalt binder and exposed polymersARMA: granules provide the colouring and also protect the shingle's asphalt from degradation that would occur from prolonged exposure to UV light. Once granules are gone in patches, the shield is gone with them.
What heat attacks, chemically: Oxidation and volatilisation of asphalt hydrocarbonsRose and TenWolde, ASHRAE Journal 2002, describing the published rationale for venting: cooler shingles slow the rate of embrittlement by slowing those two reactions. Rates rise with temperature.
What thermal cycling attacks, mechanically: Fasteners, clips, seams, laps, and long panel runsMovement is cumulative along a panel. One manufacturer's design manual states that thermal contraction elongates fastener holes in through-fastened panels by approximately 1/8 in. in extreme situations, and that restrained movement is enough to buckle and tear metal and shear fasteners.
A cool roof, in one line: High solar reflectance and high thermal emittanceDOE: both properties are measured 0 to 1, and the higher the values, the cooler the roof. The Solar Reflectance Index (SRI) rolls the two into one number, typically 0 to 100.
Where cool roofing pays: Almost always in climate zones 1A–3B; an energy penalty in zones 7 and aboveDOE BASC's zone-by-zone summary. Zones 3C, 4A and 4B: beneficial but reduced. Zone 4C: nearly zero savings over a year. Zones 5–6: usually slightly beneficial for older air-conditioned homes.
The heating penalty: Real, and stated by DOE rather than hiddenDOE: a cool roof may increase the need for heating energy in winter by reflecting sunlight that would otherwise warm the building. Even in mixed climates, however, winter penalties can be smaller than summer savings.
Ventilation's share of shingle temperature: Small. Colour and orientation are larger.Rose and TenWolde: venting a black-shingle roof had a 2–3% cooling effect on the shingles, while colour was worth 20–30%. Building Science Corporation puts colour and orientation at double or triple the durability effect of venting or not venting.
What a CRRC rating is: A measurement, not a pass markThe Cool Roof Rating Council states that it does not require rated products to meet or exceed a minimum value, and that it does not certify products, because certification implies a product meets or exceeds a standard.

Tradeoffs

This page's advice — reflectance is worth buying in the sun belt — and where that advice is wrongSection link

The position here is that in a cooling-dominated climate a reflective surface is one of the cheapest useful things you can change about a roof, and that in a heating-dominated one it is close to free at best and a mild penalty at worst. There are several honest exceptions, and one of them is large.

Best when

The roof is being replaced anyway. DOE is blunt that it is rarely economical to replace a mechanically sound roof just to increase its solar reflectance — the case for reflectance is a case about which product to buy, not about buying earlier.
You are choosing between asphalt shingle colours. DOE says cool asphalt shingles cost about the same as conventional asphalt shingles, which makes this the only decision on the page with no incremental cost to weigh.
There are ducts or an air handler in the attic. In the 1998 DOE-2.1E modelling of a new Miami house with an R-19 ceiling, moving the ducts out of the attic cut the cool-roof saving from 14% to 8% on the same prototype — roughly half the benefit was about not running cold air through a hot attic.
The ceiling insulation is thin. The same modelling put a Miami house with no ceiling insulation at a 26% cooling saving; the better the ceiling, the less work the roof surface is doing.
The building is low-slope. DOE notes that cool products are generally economical on low-slope roofs for commercial or industrial buildings, and BASC's three-year aged SRI figures for white membranes run well into the 90s — the highest numbers available on any roof.
The roof is a large, unshaded plane with a long summer. Absolute savings scale with area and with sun, which is why the same choice is trivial on a small shaded bungalow and material on a warehouse.

Think twice if

You are in climate zone 7 or above. BASC states plainly that cool roofs and walls will generally have an overall energy penalty there. In zone 4C, its finding is nearly zero savings over a year — you would be buying a colour, not an outcome.
The building is cold-climate and low-slope. Research on mechanically attached membranes in cold-climate zones finds that a reflective surface, by giving up daytime solar gain, can raise the risk of condensation in the layers under the membrane compared with a darker surface. That is a hygrothermal design question, not a colour preference.
Someone is quoting you a saving without naming your climate zone, your ceiling R-value, and where your ducts run. Those three variables move the answer by a factor of three in published modelling, so a number produced without them is not an estimate.
The roof already has R-38 or better at the ceiling, or no ducts in the attic. Under California's previous energy code edition, either one on its own exempted a steep-slope re-roof from the cool-roof requirement in the named climate zones — a regulator accepting that the reflective surface is doing less once they are true. That edition is superseded, so this is the shape of the reasoning, not a current requirement anywhere.
The covering is a heavyweight tile or slate on battens. The mechanism is different: a ventilated air space under the units already breaks much of the conductive path, and reflectance of the visible surface is not carrying the same load.
You care about appearance and the roof is visible. This is a legitimate reason and it does not need dressing up. DOE notes white roofs are coolest but cool colours are a popular alternative for roofs that can be seen by neighbours, and the aged SRI gap between them is large.
The building is historic, or in an HOA or design-review district. Colour is frequently the one thing you are not free to choose, and finding that out after signing is expensive.

What changes the answer

Climate zone — the single largest variable, and the one BASC organises its whole recommendation around.
Ceiling or roof-deck insulation level. Reflectance and insulation are partly substitutes; the more you have of one, the less the other buys.
Where the heating and cooling equipment and ducts live. Attic ducts convert attic air temperature directly into an energy bill.
Roof type. Aged SRI ceilings differ enormously by material — BASC gives roughly 20 to 35 for cool asphalt shingles against 85 or 90 for white metal and well into the 90s for white membranes.
Aged rather than initial reflectance. BASC notes three-year aged values are the industry standard and what many codes require, and DOE says savings should be based on long-term values because shingles lose some of their ability to reflect over time.
Whether the assembly is vented or a correctly designed unvented one. Both are legitimate; they put the insulation and the air control layer in different places and change what the deck temperature is doing.
Orientation and shading. Roof orientation is named alongside colour and location as a stronger driver of shingle temperature than ventilation.
Your fuel prices and equipment efficiency. The heating penalty and the cooling saving are paid in different fuels at different prices, so the sign of the net answer can flip on a tariff.
The adopted energy code where you live, and its current edition — which is not the edition that was current when the contractor learned it.

The mechanism

One surface, four things happening to itSection link

Everything on this page follows from a single fact: a roof surface in the sun is far hotter than the air around it, and it does that again tomorrow.

Section through the top of a roof assembly. The sun’s energy splits three ways at the surface — reflected, re-radiated, or absorbed and conducted downward — and two separate ageing processes run on what is left. Numbers key to the list below. Schematic, not to scale, and not a construction detail.Original diagram, Understanding Roofing.

Shortwave energy arrives. Sunlight reaches the covering. Nothing about the roof changes how much arrives; that is set by latitude, season, time of day, cloud, and which way the plane faces.
Some is reflected straight back. The fraction bounced away is the surface’s solar reflectance, measured 0 to 1. It never becomes heat in the building.
Some of the absorbed heat radiates back out. How readily a surface sheds absorbed heat as long-wave infrared is its thermal emittance, also 0 to 1. A surface can be reflective and still run hot if it emits poorly, which is why the two numbers are always quoted together and why the Solar Reflectance Index exists to combine them.
Ultraviolet light works on what is exposed. UV is a small slice of the arriving energy and does damage out of all proportion to its share, because it is energetic enough to break bonds in the organic materials at the surface.
The whole surface expands, then contracts. Up in the morning, down overnight, every day the sun shines.
What is left conducts downward. Through the covering, the underlayment, and the deck into the attic, where it raises attic air temperature and heats anything stored or ducted up there before it ever reaches the ceiling of the room below.

Why the surface is so much hotter than the forecast

A weather forecast is an air temperature measured in shade, at about 1.5 metres, in a ventilated screen. A roof is none of those things. The EPA’s heat-island work states the gap directly: on a hot, sunny summer day the sun can heat dry, exposed urban surfaces like roofs and pavement to temperatures 50 to 90 °F hotter than the air. DOE’s Building America Solution Center puts the absolute figure at 150 °F or higher for standard or dark roofs in hot climates, and illustrates a Florida house whose standard asphalt shingle roof heats to 140 °F.

Two calibrations on that 50-to-90 range before it gets used. It is a statement about dry exposed surfaces in general, not about roofing in particular; and EPA’s own roofing-specific statement elsewhere is narrower — conventional roofing materials reaching as much as 66 °F warmer than the surrounding air on a warm day. Treat 50 to 90 as the wide end of the published range rather than as what your roof does, and treat every surface figure on this page as an illustration of magnitude.

That gap is the whole hazard. A 100 °F afternoon is not a 100 °F roof; it is plausibly a 150 °F to 190 °F roof, and every rate that matters — chemical reaction, softening, outgassing — responds to the surface temperature, not the reported one.

UV: the shield, and what happens when it goes

On an asphalt shingle the granules are the sunscreen. The Asphalt Roofing Manufacturers Association states it plainly: granules provide the colouring to shingles, and they also protect the shingle’s asphalt from degradation that would occur from prolonged exposure to UV light. The same document says the primary purpose of the asphalt is to shed water and to hold the granules in place. Those two sentences together describe a system where the part doing the waterproofing depends on the part doing the shielding staying put.

So granule loss is not a cosmetic finding on an older roof; it is the moment the clock speeds up. Bare patches expose asphalt directly to UV, the exposed asphalt embrittles faster, embrittled asphalt releases more granules, and the process feeds itself. What you can see from the ground is a slope that has gone matte and slightly darker, and grit accumulating where the downspout discharges.

Membranes and polymer coverings have the same problem in a different wrapper. A single-ply membrane, a coating, an exposed sealant, a plastic pipe boot, the rubber gasket under an exposed fastener — every one of them is an organic polymer with a UV stabiliser package that is consumed over time. The small ones fail first, which is why boots and gaskets are the classic ten-year maintenance item on a thirty-year roof.

Heat: the chemistry, which is separate from the UV

Rose and TenWolde, writing in the ASHRAE Journal in 2002, describe the published rationale for ventilating shingle roofs: venting cools shingles and thereby affects the rate of embrittlement by reducing the rates of oxidation and volatilisation of asphalt hydrocarbons. Read that as a statement about what heat does rather than about what venting does. Asphalt is a mixture; the lighter, more mobile fractions that keep it flexible are driven off over time, and oxygen attacks what remains. Both processes go faster hot.

That is why heat exposure is best thought of as a dose rather than an event, and why the sensible unit is not “how hot did it get once” but “how many hours has this surface spent above some temperature over twenty years.”

Thermal cycling: the mechanics, which is separate again

Everything expands when heated. A roof plane heats and cools once a day, and the movement accumulates along the length of a continuous element. On a shingle roof it shows up as the slow working of nails and the gradual opening of thermal splits in aged material. On a metal roof it is a design load: movement is cumulative along a panel run, which is why standing seam uses sliding clips and why through-fastened panels elongate their own fastener holes instead. On a low-slope membrane it concentrates at seams, at terminations, and over deck joints.

The counter-intuitive part, worked through in the two-cities comparison below, is that the daily air-temperature swing is remarkably similar in hot and mild climates. That is not the same thing as the covering’s own daily swing, which is set mostly by solar gain and is not something a weather normal measures. What extreme heat plainly does change is the peak the material reaches and the number of hours it spends near it — which is what drives the chemistry.

Original analysis

Two cities, one shingle: what actually differs between Phoenix and SeattleSection link

The intuitive story is that a hot climate swings the roof harder. The normals do not support that. What the hot climate changes is the peak the surface reaches and how long it stays there — which is a chemistry story, not a mechanics one.

Phoenix and Seattle compared on NOAA 1991–2020 monthly climate normals, with an illustrative dark-roof surface band derived by adding the EPA's published 50–90 °F surface-over-air range to the normal daily maximum. Airport stations; normals, not extremes.
Measure	Phoenix Sky Harbor, AZ	Seattle–Tacoma, WA	What it changes on the roof
Warmest month, normal daily maximum	106.5 °F (July)	77.6 °F (August)	The peak the covering has to survive. Add the EPA band and the Phoenix surface lands roughly between 157 °F and 197 °F; the Seattle surface between roughly 128 °F and 168 °F.
Coldest month, normal daily minimum	45.3 °F (December)	37.1 °F (December)	The bottom of the annual range. Both are monthly averages of daily minima rather than the coldest night either city sees, so they are not evidence that frost never occurs — but in neither city is a typical winter the freeze-thaw story told on a different page.
Annual span (coldest normal minimum to warmest normal maximum)	61.2 °F	40.5 °F	The seasonal stroke of the assembly. Phoenix is about half again as wide, which matters most to long continuous elements like metal panel runs.
Daily span in the warmest month, air temperature	22.0 °F (July: 106.5 / 84.5)	20.4 °F (August: 77.6 / 57.2)	Almost identical — as air. The covering's own daily excursion is larger than this and is driven mostly by solar gain, which a temperature normal does not report. What this row shows is that the climate's daily cycle is not what separates the two cities, so “thermal cycling” on its own does not explain the difference in service life.
Months whose normal daily maximum is 90 °F or higher	Five — May, June, July, August, September	None, in any month	Dose, not amplitude. Phoenix spends five months of a normal year at a daily peak Seattle never reaches, and reaction rates respond to temperature, not to the size of the swing.
Difference at the summer peak	+28.9 °F against Seattle	—	Same product, same colour, same slope: the Phoenix surface sits about 29 °F hotter at the annual peak, all day, for five months, for the life of the roof.

Read this table one item at a time

Warmest month, normal daily maximum

Phoenix Sky Harbor, AZ: 106.5 °F (July)
Seattle–Tacoma, WA: 77.6 °F (August)
What it changes on the roof: The peak the covering has to survive. Add the EPA band and the Phoenix surface lands roughly between 157 °F and 197 °F; the Seattle surface between roughly 128 °F and 168 °F.

Coldest month, normal daily minimum

Phoenix Sky Harbor, AZ: 45.3 °F (December)
Seattle–Tacoma, WA: 37.1 °F (December)
What it changes on the roof: The bottom of the annual range. Both are monthly averages of daily minima rather than the coldest night either city sees, so they are not evidence that frost never occurs — but in neither city is a typical winter the freeze-thaw story told on a different page.

Annual span (coldest normal minimum to warmest normal maximum)

Phoenix Sky Harbor, AZ: 61.2 °F
Seattle–Tacoma, WA: 40.5 °F
What it changes on the roof: The seasonal stroke of the assembly. Phoenix is about half again as wide, which matters most to long continuous elements like metal panel runs.

Daily span in the warmest month, air temperature

Phoenix Sky Harbor, AZ: 22.0 °F (July: 106.5 / 84.5)
Seattle–Tacoma, WA: 20.4 °F (August: 77.6 / 57.2)
What it changes on the roof: Almost identical — as air. The covering's own daily excursion is larger than this and is driven mostly by solar gain, which a temperature normal does not report. What this row shows is that the climate's daily cycle is not what separates the two cities, so “thermal cycling” on its own does not explain the difference in service life.

Months whose normal daily maximum is 90 °F or higher

Phoenix Sky Harbor, AZ: Five — May, June, July, August, September
Seattle–Tacoma, WA: None, in any month
What it changes on the roof: Dose, not amplitude. Phoenix spends five months of a normal year at a daily peak Seattle never reaches, and reaction rates respond to temperature, not to the size of the swing.

Difference at the summer peak

Phoenix Sky Harbor, AZ: +28.9 °F against Seattle
Seattle–Tacoma, WA: —
What it changes on the roof: Same product, same colour, same slope: the Phoenix surface sits about 29 °F hotter at the annual peak, all day, for five months, for the life of the roof.

Method: NOAA National Centers for Environmental Information 1991–2020 monthly climate normals for stations USW00023183 (Phoenix Sky Harbor) and USW00024233 (Seattle–Tacoma). The surface band is the normal daily maximum plus the EPA's published 50–90 °F surface-over-air range for dry exposed surfaces in full sun; it is an illustration of magnitude, not a measurement of any roof, and EPA's own roofing-specific figure elsewhere is narrower. Every span in this table is air temperature; the covering's own temperature swing is larger and depends on solar gain, which these normals do not report. Normals are thirty-year averages of daily values, so real days run both hotter and colder than any figure here. The comparison holds colour, slope, orientation, shading, and assembly constant — no two real houses do.

What follows from this, and what does not

It does follow that the temperature-driven chemistry — oxidation and volatilisation of the binder, consumption of UV stabiliser packages, softening and creep of adhesives — runs far harder in Phoenix. Rose and TenWolde’s summary of Cash and Lyon says the same thing from the other direction: shingle temperature is more strongly a function of geographic location, roof orientation and surface colour than of ventilation.

It does not follow that Seattle roofs live easy lives. They fail for other reasons on a different page — sustained wetting, biological growth, and a marine climate — and a covering that reaches the top of its planning range for heat may still be replaced early for moss. The claim here is narrow: heat and UV dose is what this axis measures, and on this axis the two cities are not close.

It also does not follow that the difference is uniform across a single roof. Orientation is named in the same research as a driver on the order of colour, which means a south-west-facing plane and a north-facing plane on the same house in the same city age at meaningfully different rates. That is why a competent inspection report reports slope by slope rather than giving a roof one age.

Worked example

How far a metal panel actually moves, and what has to absorb itSection link

Thermal movement is the one part of this page that can be calculated rather than described. The arithmetic is simple; the point is the magnitude, and what happens if nothing in the design is allowed to slide.

Linear thermal movement over a 40 ft (480 in.) panel run, using coefficients of expansion published in one manufacturer's design manual. Product-specific figures used arithmetically, not a design calculation.
Case	Coefficient (in./in. per 100 °F)	Temperature differential	Movement over a 40 ft run
Light-coloured steel panel, uninsulated building	.00065	100 °F — the minimum the manual says to design for in this case	480 × .00065 × 1.00 = 0.31 in., about 5/16 in.
Dark steel panel, insulated building, Phoenix	.00065	135 °F — 180 °F exposed metal down to a 45.3 °F December normal daily minimum	480 × .00065 × 1.35 = 0.42 in., about 7/16 in.
Dark aluminium panel, insulated building, Phoenix	.00128	135 °F, as above	480 × .00128 × 1.35 = 0.83 in., about 13/16 in.

Read this table one item at a time

Light-coloured steel panel, uninsulated building

Coefficient (in./in. per 100 °F): .00065
Temperature differential: 100 °F — the minimum the manual says to design for in this case
Movement over a 40 ft run: 480 × .00065 × 1.00 = 0.31 in., about 5/16 in.

Dark steel panel, insulated building, Phoenix

Coefficient (in./in. per 100 °F): .00065
Temperature differential: 135 °F — 180 °F exposed metal down to a 45.3 °F December normal daily minimum
Movement over a 40 ft run: 480 × .00065 × 1.35 = 0.42 in., about 7/16 in.

Dark aluminium panel, insulated building, Phoenix

Coefficient (in./in. per 100 °F): .00128
Temperature differential: 135 °F, as above
Movement over a 40 ft run: 480 × .00128 × 1.35 = 0.83 in., about 13/16 in.

Arithmetic: run length in inches × coefficient × (ΔT ÷ 100). The coefficients, the 100 °F minimum design differential, and the statement that a dark panel on an insulated building can reach 180 °F exposed metal temperature are all from one manufacturer's 2011 design manual and are product-specific. The 45.3 °F is the NOAA 1991–2020 December normal daily minimum for Phoenix Sky Harbor — a normal, not a record low. The manual itself says to use the lowest expected winter temperature, which is colder than a normal daily minimum, so a real design differential would be larger than the one worked here. Panel length, fixed-point location, clip type, substrate and fastener pattern for a real roof are set by the manufacturer's instructions and by a qualified designer, not by this table.

Where that half-inch goes

Half an inch sounds trivial until you ask what absorbs it. In a standing seam system, sliding clips let the panel move relative to the structure while the seam stays closed; that is the entire reason the system costs more than a through-fastened one, and it is why the standing seam guide spends so long on clips.

In a through-fastened panel, nothing slides. The panel is pinned at every screw, so the movement is taken up by the screw holes themselves. The same manual states that thermal contraction will elongate those holes by approximately 1/8 in. in extreme situations, and that a properly installed fastener with an adequate sealing washer should prevent that becoming a leak. That is an honest description of a system whose watertightness depends on a rubber washer surviving a hole that is slowly getting bigger. It is not a defect; it is the design, and it is why re-fastening appears on maintenance schedules for this panel type and not for standing seam.

And if movement is restrained — a panel fixed at both ends, a curb or penetration that pins the middle of a run, a rooftop unit bolted through — the manual is blunt: because the movement is cumulative, the force is sufficient to buckle and tear metal and shear fasteners. Most of the ugly metal-roof failures that get blamed on “cheap panels” are this.

The last number is the one that puts the scale of it in perspective. The same manufacturer cycled a panel-and-clip assembly 100,000 times to simulate up to five cycles a day for 50 years, then measured the contact points with a micrometer: a maximum of 0.005 in. of material lost. A properly detailed metal roof is not fighting thermal cycling. It is accommodating it, which is a different thing, and the failures happen where the accommodation was left out.

Worked example

Does a cool roof pay? Four houses, one intervention, four answersSection link

This is the section the page exists for. Every row is the same change — raising the roof's solar reflectance — applied to a different building, using published results rather than an estimate.

Published results for raising roof solar reflectance, from Parker et al., ASHRAE Transactions 1998. Miami and Detroit cases are DOE-2.1E modelling of prototype buildings; the Florida field measurements are separate and are noted in the text.
The building	What changed	Published result	What it demonstrates
Miami, no ceiling insulation	Reflective roof against a dark base case	26% cooling energy saving	The worse the ceiling, the more work the roof surface is doing. Reflectance is partly a substitute for insulation you do not have.
Miami new-construction prototype, R-19 ceiling, ducts in the attic	The same change	14% cooling saving — 730 kWh a year	Insulate the ceiling and the same roof buys you a lot less. Nothing about the roof changed; the building around it did.
Miami new-construction prototype, R-19 ceiling, ducts inside the conditioned space	The same change	8% cooling saving — 362 kWh a year	Move the ducts out of the attic and the saving halves again. Roughly half of this cool roof's value was never about the ceiling at all — it was about not running cold air through a hot attic.
Detroit, existing house	The same change	Annual space-conditioning cost $0.508/ft² before, $0.504/ft² after	Cooling fell, heating rose, and the two very nearly cancelled. This is what “the heating penalty is real” looks like when it is a number instead of a warning.

Read this table one item at a time

Miami, no ceiling insulation

What changed: Reflective roof against a dark base case
Published result: 26% cooling energy saving
What it demonstrates: The worse the ceiling, the more work the roof surface is doing. Reflectance is partly a substitute for insulation you do not have.

Miami new-construction prototype, R-19 ceiling, ducts in the attic

What changed: The same change
Published result: 14% cooling saving — 730 kWh a year
What it demonstrates: Insulate the ceiling and the same roof buys you a lot less. Nothing about the roof changed; the building around it did.

Miami new-construction prototype, R-19 ceiling, ducts inside the conditioned space

What changed: The same change
Published result: 8% cooling saving — 362 kWh a year
What it demonstrates: Move the ducts out of the attic and the saving halves again. Roughly half of this cool roof's value was never about the ceiling at all — it was about not running cold air through a hot attic.

Detroit, existing house

What changed: The same change
Published result: Annual space-conditioning cost $0.508/ft² before, $0.504/ft² after
What it demonstrates: Cooling fell, heating rose, and the two very nearly cancelled. This is what “the heating penalty is real” looks like when it is a number instead of a warning.

From Parker, Huang, Konopacki, Gartland, Sherwin and Gu, ASHRAE Transactions Vol. 104 Pt. 1 (1998). The dollar figures use 1996 national average energy prices ($0.084/kWh, $0.63/therm) and prototype buildings of that era; they are obsolete as money and used here only as a ratio. Percentages are cooling energy, not total energy and not total bill. Separately, the same paper reports measured air-conditioning savings on 11 Florida homes whitened mid-summer averaging 19% and ranging from 2% to 43%.

The paper’s own summary of where the line falls

The authors state it directly: except in the northernmost locations — Minneapolis and Detroit — and cool, cloudy locations — Seattle and San Francisco — the combined cost of heating and cooling was lower for reflective roof surfaces than for conventional ones. They also found that where heating dominates, above about 40 degrees north latitude, there was little advantage from reflective roofing, and in the coldest locations added insulation was clearly the better place to put the money.

The heating penalty itself, in existing housing, ranged from 3 percent in Miami to 6 percent in San Francisco. Small numbers — but small numbers applied to a large heating bill and set against a saving on a small cooling bill is exactly how a positive result turns neutral.

The convergence worth noticing

Twenty-five years later, California’s energy code — in the 2022 edition, now superseded — exempted a steep-slope re-roof from its cool-roof requirement when the ceiling carried at least R-38, or when there were no ducts in the attic in several named climate zones. Those are the same two variables that moved the Miami result from 26 percent to 14 percent to 8 percent.

A regulator with a compliance burden to justify and a research team with a simulation model arrived independently at the same two questions. That is a stronger signal than either on its own, and it is the practical test to apply to any cool-roof proposal: what is my ceiling R-value, and where are my ducts? If the person selling you the roof cannot answer those, the saving they quoted is not about your house.

The ceiling on reflectance

How cool a roof can get depends more on what it is than on what colour it isSection link

Solar Reflectance Index rolls reflectance and emittance into one number, typically 0 to 100, higher being cooler. The spread between materials is far larger than the spread between colours within a material.

Three-year aged Solar Reflectance Index ranges by roof type, as published by the DOE Building America Solution Center. Aged rather than initial values, because aged is what codes and standards generally require.
Roof type	Typical three-year aged SRI	What sets the ceiling	Where it applies
White or light metal finish	As high as 85 or 90	A smooth, factory-applied coating on a non-porous substrate. Parker et al. add that smooth metal roofs seem to hold their reflectance best as they weather — but they label that observation anecdotal, so treat it as a reason to ask for the aged number rather than as a substitute for it.	Steep-slope and low-slope. The highest steep-slope reflectance generally available.
Dark metal finish (brown, red)	Above 30	Visible colour, but the coating can still be engineered to reflect near-infrared — most of the solar energy is not visible light.	Steep-slope. The compromise where appearance is constrained.
Cool asphalt shingle	Generally about 20 to 35	The granules sit in a dark asphalt substrate that shows through. Laboratory testing of 14 shingle colours found reflectances of only 5–25%, and a deliberately white shingle reached only 31%.	Steep-slope. Costs about the same as a conventional shingle, which makes it the cheapest reflectance available.
White single-ply membrane	Well into the 90s	A homogeneous white polymer surface with nothing dark underneath it.	Low-slope only. This is why the strongest cool-roof case in the country is a large flat commercial roof in the south.
Dark membrane	About 5 to 20	A black or dark surface absorbing nearly everything that hits it.	Low-slope. Still specified deliberately in cold climates, where the winter gain is worth having and the condensation risk is lower.

Read this table one item at a time

White or light metal finish

Typical three-year aged SRI: As high as 85 or 90
What sets the ceiling: A smooth, factory-applied coating on a non-porous substrate. Parker et al. add that smooth metal roofs seem to hold their reflectance best as they weather — but they label that observation anecdotal, so treat it as a reason to ask for the aged number rather than as a substitute for it.
Where it applies: Steep-slope and low-slope. The highest steep-slope reflectance generally available.

Dark metal finish (brown, red)

Typical three-year aged SRI: Above 30
What sets the ceiling: Visible colour, but the coating can still be engineered to reflect near-infrared — most of the solar energy is not visible light.
Where it applies: Steep-slope. The compromise where appearance is constrained.

Cool asphalt shingle

Typical three-year aged SRI: Generally about 20 to 35
What sets the ceiling: The granules sit in a dark asphalt substrate that shows through. Laboratory testing of 14 shingle colours found reflectances of only 5–25%, and a deliberately white shingle reached only 31%.
Where it applies: Steep-slope. Costs about the same as a conventional shingle, which makes it the cheapest reflectance available.

White single-ply membrane

Typical three-year aged SRI: Well into the 90s
What sets the ceiling: A homogeneous white polymer surface with nothing dark underneath it.
Where it applies: Low-slope only. This is why the strongest cool-roof case in the country is a large flat commercial roof in the south.

Dark membrane

Typical three-year aged SRI: About 5 to 20
What sets the ceiling: A black or dark surface absorbing nearly everything that hits it.
Where it applies: Low-slope. Still specified deliberately in cold climates, where the winter gain is worth having and the condensation risk is lower.

SRI ranges from the DOE Building America Solution Center; the shingle reflectance testing is from Parker et al. (1998) and is laboratory solar reflectance rather than SRI, so the two are quoted separately rather than combined. SRI is not a performance guarantee, and a CRRC rating is a measurement rather than a pass mark. Whether any of these numbers satisfies a requirement is a question for the adopted code in your jurisdiction.

The practical reading is that if reflectance genuinely matters to you, it is a material decision before it is a colour decision. Moving from a dark shingle to a cool shingle is free and gains you something. Moving from a shingle to a white metal or a white membrane gains you two to four times as much — and costs considerably more, changes the look of the building, and brings an entirely different set of tradeoffs that the metal-against-asphalt comparison works through properly.

One caution about aged numbers in general. DOE lists the incremental cost of keeping a cool roof clean and reflective as one of three factors that decide whether it pays. Three-year aged ratings already include normal soiling from the CRRC’s outdoor weathering, so buying on the aged number is the honest way to buy. Buying on the initial number is buying the day the roof was installed.

Cost and lifecycle

What reflectance is worth, and what it costsSection link

There is no dollar figure on this page for a roof. There are published percentage savings, and one published statement about the price difference between a cool shingle and a conventional one.

Cool asphalt shingle, price against conventional: About the sameDOE, Consumer Guide to Cool Roofs. This is the reason the colour decision inside asphalt is the cheapest one on the page.
Annual air-conditioning energy, single-storey building: Up to 15% lowerDOE, for substituting a cool roof for a conventional roof in warm or hot climates.
Cooling costs, cool-colour roof on a home: At least 5% to 20% lowerDOE. Savings vary and are greatest in hot and warm climates.
Measured, 11 Florida homes whitened mid-summer: 19% average, ranging 2% to 43%Parker et al., ASHRAE Transactions 1998. The spread is the point: same intervention, same state, one order of magnitude between the best and worst house.
Winter heating energy: IncreasesDOE: a cool roof may increase the need for heating energy in winter by reflecting sunlight that would otherwise warm the building.

Units: Percentage change in annual space-conditioning energy or cost, against a conventional darker roof on the same building
Scope included: The energy effect of changing the roof surface only — solar reflectance and thermal emittance — with the assembly, insulation, and equipment otherwise unchanged
Not included: Installed price of any roof, regional labour and material rates, incentives, roof replacement itself, structural work, and any cost of keeping a reflective surface clean
Geography: United States. The DOE figures are national statements for warm and hot climates; the 19% measurement is 11 houses in central and south Florida
Data as of: DOE consumer guide dated August 2021 (DOE/EE-2478); Florida field measurements published in ASHRAE Transactions in 1998
Confidence: Directional only. The published ranges are wide because the answer genuinely varies by climate zone, insulation, duct location, and roof type — the four variables the worked example below moves one at a time.
Method: How this figure is built

The 1998 measurements are old, and the prices behind the modelling in that same paper are 1996 national averages for electricity and natural gas. That matters for the dollars and not much for the physics: the ratios between the cases are what this page uses, and the current DOE and BASC guidance points the same direction. Where the age of the data does matter, it is said so in place.

One structural warning about quotes. A cool roof is sold on an energy saving, and an energy saving is the easiest number in roofing to state without a basis. If a proposal contains a percentage or a dollar figure, ask which climate zone, which ceiling R-value, and which duct location it assumes, and compare it against the other proposals on the same terms. A saving quoted without those is not a forecast about your house.

DOE's own guidance draws the line for you: it is rarely economical to replace a mechanically sound roof just to increase its solar reflectance. The decision on this page is which surface to buy when the roof is being replaced anyway — not whether to replace it sooner.

Considerations

What changes this on a real buildingSection link

Climate

BASC organises its cool-roof recommendation by climate zone, and the shape of that recommendation is the honest answer to “does this pay?” In zones 1A through 3B, cool roofs and walls are almost always beneficial. In 3C, 4A and 4B they are usually beneficial but the benefits are reduced by the proportional increase in heating energy. In 4C they have been shown to have nearly zero savings over the course of a year. In zones 5 and 6 they are usually slightly beneficial for older air-conditioned homes. In zone 7 and above they will generally have an overall energy penalty.

The reason the penalty is smaller than intuition suggests is geometric. BASC notes that the solar impact on roof and wall temperature is lower in winter than in summer, and the EPA adds the rest of the list: lower winter sun angles, shorter days, insulation, and snow cover sitting over the roof anyway.

Climate zone is an IECC designation for a county, not a description of your microclimate, your shading, or your orientation. A zone tells you which way the answer probably leans; it does not compute it.

Code and jurisdiction

There is no nationwide building or energy code for site-built construction in the United States, and there is no national cool-roof requirement. Some jurisdictions require reflectance on some roofs; most do not. California is the best-known example, and it is worth walking through precisely because it shows how much a real rule depends on details a national page cannot know.

California’s energy code is the Building Energy Efficiency Standards, Title 24 Part 6, published by the California Energy Commission. The current edition is the 2025 Energy Code, and the Commission states that buildings whose permit applications are applied for on or after 1 January 2026 must comply with it. This page does not publish the 2025 numeric thresholds, because we have not read the adopted text; the Cool Roof Rating Council’s summary points to sections 150.1(c)11 and 150.2(b)1I for the prescriptive residential values, and compliance is shown with a CRRC product label.

The previous edition — the 2022 Energy Code, which governed permits applied for on or after 1 January 2023 and is now superseded — is quoted here for one reason only: its exemption list is a building-science argument written as law. Under a public-agency permit guide to that edition, a steep-slope re-roof did not need a cool roof if the ceiling carried at least R-38, or if there were no ducts in the attic in several named climate zones, or if there was a qualifying attic radiant barrier, or if there was at least R-2 continuous insulation above or below the deck. Cool-roof requirements were triggered only when 50 percent or more of the roof was being replaced.

In other words, a jurisdiction that had actually thought about this stopped requiring the reflective surface exactly where the 1998 modelling said its value collapses. That convergence is the strongest argument on this page, and it is also the clearest demonstration of why the numbers are local: those thresholds were current for three years and are not current now.

Record the jurisdiction, the adopted edition, the amendments, the effective date, and the official URL for any code claim, and confirm it with the authority having jurisdiction. The California figures above describe a superseded edition and are printed to show the shape of such a rule, not to state a requirement anywhere — including in California.

Moisture and ventilation

A reflective roof is a colder roof, and a colder roof is a wetter one in some assemblies. This is the real downside that cool-roof marketing does not mention. Research presented to IIBEC on mechanically attached membrane roofs in cold-climate zones found that a white exterior surface applied to cut cooling loads increases condensation risk, because the assembly gets less solar energy during daylight hours to drive moisture back out, while flat roofs already have a high potential for night-time overcooling. Across their 128 simulated cases the cool roof accumulated roughly twice the moisture below the membrane that a black roof did.

Read the rest of that paper before treating it as an argument against white membranes, though. The two variables that decided whether a case came out safe or failing were the indoor moisture supply and the rate of air intrusion under the membrane; of the four things the authors varied, the climate itself had the least influence across the zones 4 to 7 they modelled. The surface colour moves the risk. It is not what sets it.

This is not a reason to avoid reflectance in a hot climate, where it does not arise. It is a reason that a low-slope reflective roof in a cold or mixed climate is a hygrothermal design question — vapour retarder placement, insulation above the deck, air-tightness of the deck — answered by someone modelling that assembly, not by a colour selection.

The hot-humid version of the problem lives on the high-humidity page, and the cold version, where water in a porous material freezes and expands, is a separate hazard entirely.

Ventilation, and unvented assemblies

Ventilation is usually sold as the thing that keeps shingles cool. The research says it is a minor contributor. Rose and TenWolde report that ventilating a black-shingle, truss-framed roof assembly in Illinois had a 2 to 3 percent cooling effect on the shingles, while the effect of colour was 20 to 30 percent, and that modelling by Cash and Lyon found shingle temperature is more strongly a function of geographic location, the direction the roof faces, and surface colour than of ventilation. Their summary sentence is worth quoting exactly: “Venting cools shingles, but the cooling effect is not strong.”

That does not make ventilation pointless. It has other jobs — moisture control in cold climates, ice-dam control, attic conditions — and the ventilation guide works through net free area and the intake and exhaust balance properly. It does mean that “we will add more vents so your shingles last longer” is a weak claim sitting next to a strong one about colour.

A correctly designed unvented assembly is equally legitimate, and the heat penalty for choosing one is quantified rather than hypothetical. Building Science Corporation states that shingles on unvented attic assemblies operate at a slightly higher temperature — a 2 or 3 °F rise in average temperature for asphalt shingles and a corresponding 10 °F rise for the sheathing — and that applying the Arrhenius equation, a 10 percent reduction in useful service life should be expected, comparable to the effect of installing radiant barriers. It then adds the line this whole page turns on: the colour of shingles and roof orientation have a more profound effect on shingle durability than the choice of venting or not venting, by double or triple.

There is a further reason to think about vents that has nothing to do with temperature. Rose and TenWolde note that roof vents may play a detrimental role in forest fires in residential areas, because vents admit sparks and burning brands. In a wildland-urban interface that consideration outranks the thermal one — see the wildfire page.

There is no universal ventilation ratio, and “more ventilation is always better” is false. Vented and correctly designed unvented assemblies are both legitimate. Which applies to your building, and what net free area is required, is set by the adopted code edition, the climate zone, and the assembly — confirm with the authority having jurisdiction and a qualified designer.

Colour, and the ceiling asphalt puts on it

Colour is the strongest lever a homeowner controls, and on asphalt shingles it is also the most limited. Laboratory testing of 14 shingle colours reported in the 1998 Florida work found solar reflectances of only 5 to 25 percent across the range; six of the most popular grey and brown colours averaged 10 percent; and a shingle specially manufactured to be white still tested at only 31 percent. The reason given is structural to the product: the granules are impregnated into a dark asphalt substrate, and that substrate shows through.

Modern cool-shingle technology has moved that ceiling, but not as far as the marketing implies. BASC’s three-year aged SRI figures put cool asphalt shingles generally at about 20 to 35, against 85 or 90 for white metal finishes and well into the 90s for white membranes. If reflectance is your priority, the material decision matters more than the colour decision inside a material.

Two more points on colour. First, a cool roof need not be white: BASC notes roofing products of any colour can be made to reflect near-infrared radiation, which is most of the solar energy, so colour and coolness are not the same axis. Second, reflectance ages. Older field studies cited in the 1998 work found white surfaces could lose as much as 20 percent of their reflectance in two years with degradation slowing after that, which is exactly why three-year aged ratings exist.

Slope and drainage

Slope decides which product family you are choosing from, and that decides the reflectance ceiling. DOE separates the market at a pitch of 2:12: below it, cool thermoplastic membranes, elastomeric coatings and metal products; above it, cool asphalt shingle, clay tile, concrete tile and metal. Slope also changes the geometry of the problem, because a low-slope roof faces the sun much more squarely at midday than a steep one does, and it is usually a larger uninterrupted plane. Both of those are why the strongest cool-roof case in the country is a large flat commercial roof in the south.

Maintenance

A reflective roof is the one roof surface whose performance depends on staying clean. DOE lists the incremental cost of keeping a cool roof clean and reflective as one of three factors affecting cost-effectiveness, alongside the incremental purchase price and the loss of reflectance over time. The industry answer to that is the three-year aged rating rather than a cleaning schedule — the CRRC weathers samples outdoors for three years, and ASTM D7897 provides a laboratory soiling-and-weathering practice used to estimate the same thing faster. Buy on the aged number and you have already priced the dirt in.

Fire

A cool roof and a fire-rated roof are different claims about different things, and a coating applied to change one can change the other. Class A, B, or C is a property of a tested assembly — deck, underlayment, and covering together — not of the visible surface alone. Adding a field-applied reflective coating to an existing roof creates an assembly nobody tested. If a fire classification matters to you, whether for code, insurance, or a wildfire exposure, the question to put in writing is which specific listed assembly the finished roof will be.

A roof's fire classification belongs to the tested assembly and to the listing that covers it, not to a covering or a coating in isolation. Verify the listing for the assembly as it will actually be built.

Access and site conditions

Every diagnostic observation on this page is available from the ground or from a document: matte, darkened slopes; grit at the downspout; a bare patch visible from an upstairs window in a neighbouring house; a photograph taken from the pavement with a zoom lens; an inspection report you commissioned. In extreme heat that is not merely the preferred route, it is the only sane one, and a contractor who proposes a midday walk-through in a heat warning is telling you something about how they run a crew.

Do not go onto a roof to look for heat or UV damage. There is no urgency here that justifies it — nothing on this page is failing this afternoon that will not be equally visible in October.

Warranty and repair

What a reflectance number promises, and what it does notSection link

Heat and UV touch four separate documents, and only one of them is about water.

The covering warranty: A shingle or membrane warranty is about the product doing its job for a period, with a proration schedule and an exclusion list. Heat and UV ageing is the ordinary path by which a covering reaches the end of that period, so “it wore out in the sun” is generally the covered life running out rather than a defect. Read the exclusions for language about ambient temperature, ventilation, and colour, and read the proration table, because a prorated remedy in year eighteen of a thirty-year document is a smaller number than most people expect. See how to read a roofing warranty.
The ventilation clause: Rose and TenWolde record that most asphalt shingle manufacturers warrant their products only for ventilated roofs, and that such clauses date from the 1980s. That is a warranty condition, not a statement of physics — the same paper shows ventilation is a minor factor in shingle temperature. It still matters commercially: if you are considering an unvented assembly, the clause is a term you have to negotiate or accept with your eyes open, and it should be settled in writing before installation rather than discovered during a claim.
The reflectance rating, which is not a warranty at all: A CRRC rating is a measurement. The Council states that it does not require rated products to meet or exceed a minimum value for solar reflectance and thermal emittance, and that it does not certify products, because certification implies that a product meets or exceeds a standard. A CRRC label therefore tells you what a product measured, initially and after three years of weathering — not that it is good, not that it complies with anything, and not that it will save you money.
The coating warranty, if there is a coating: A field-applied reflective coating is a separate product with its own document, its own surface-preparation and application conditions, and its own recoat obligation. Coatings are the easiest place on a roof to void a warranty by accident, because the conditions — substrate cleanliness, moisture content, minimum and maximum application temperature, film thickness — are all invisible after the fact.

Repairability

Heat and UV damage is the least repairable kind of roof damage, because it is uniform. A hail strike or a wind loss is local and can be patched; a slope that has spent twenty years in Arizona sun has aged everywhere at once, and there is no sound part to tie into.

Early-life, localised: boots, gaskets, exposed sealants and pipe flashings fail long before the field does and are genuinely repairable. This is the maintenance that buys the back half of a covering’s planning range.
Mid-life, membrane: a low-slope membrane that is still watertight but chalking can sometimes be recoated, which restores reflectance and adds UV protection. Whether that is a genuine life extension or deferred replacement dressed up depends entirely on the condition of what is under it.
Late-life, shingle: once granule loss is widespread, a brittle shingle cannot be lifted to work under it without breaking. That is the practical point at which repair stops being available and the conversation moves to repair or replace.
Colour matching: a sun-aged roof has shifted colour. New units of the same product will not match, and on a visible slope that is a real cost of patching rather than an aesthetic quibble.

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

Every one of these is answerable from a product data sheet and a load of local experience. An installer who cannot answer them is selling colour.

What is the three-year aged solar reflectance and thermal emittance of the exact product you are proposing, and can I see the CRRC label?
Aged values are the industry standard and what many codes require; initial values flatter the product. A good answer is two numbers and a label. A weak answer talks about the colour or hands you a brochure claim with no rating behind it.
Which climate zone am I in, and what does the current adopted energy code require of this re-roof?
This is a question with a checkable answer that changes with the edition your jurisdiction has adopted. An installer working from the edition they learned five years ago will confidently give you a superseded requirement, which is how permits get failed.
Are there ducts or an air handler in my attic, and does your proposal do anything about them?
Attic duct location was worth roughly half of the modelled cool- roof saving in the Florida work. If the answer is yes and the proposal ignores it, the cheapest remaining energy improvement on the building is not on the roof at all.
What is the ceiling insulation now, and would you spend this money there instead?
An installer who will tell you that insulation beats reflectance on your particular building is worth more than one who will not. In the 1998 modelling, every location where a new reflective roof saved more than another increment of ceiling insulation lay below 37 degrees north latitude; above about 40 degrees north the authors found little advantage in reflective roofing at all, and in the coldest locations added insulation was clearly the better choice.
If you are proposing a coating, what is the substrate condition, the recoat interval, and what voids the coating warranty?
Coatings carry an ongoing obligation that the sales conversation usually skips. A specific recoat interval and a named surface- preparation standard is a real answer; “it lasts for years” is not.
What listed assembly will this roof be when you have finished, for fire classification?
Fire class belongs to a tested assembly. Any answer that names only the covering has not understood the question, and that matters most on exactly the buildings where a reflective surface is most attractive.
In what months and hours will your crew be on my roof, and what is your heat plan?
This is a safety question and a quality question at once. Work done at 2 p.m. in August on a surface DOE puts at 150 °F or higher is worse work, and sealant strips and self-adhered membranes have application-temperature limits at both ends.

Require these in writing

The exact product, and its three-year aged solar reflectance, thermal emittance, and SRI as rated
Whether the proposal is priced against the current adopted energy code edition, named by year
Ventilation scope, if any, stated as net free area of intake and of exhaust separately
Whether the assembly is vented or unvented, and what the covering warranty says about that
The listed fire-rated assembly the finished roof will constitute
For coatings: surface preparation, dry film thickness, application temperature limits, and recoat interval
Any energy-saving percentage claimed, with the climate zone, ceiling R-value, and duct location it assumes

What goes wrong

Misconceptions and failure modesSection link

Common misconceptions

Common belief
A white roof will cut my energy bill in half.
What is actually true
DOE’s published range for a cool-colour roof on a home is at least 5 to 20 percent off cooling costs, and up to 15 percent of annual air-conditioning energy for a single-storey building in a warm or hot climate. Cooling is a fraction of a whole bill, and the fraction of that fraction which the roof controls depends on the ceiling and the ducts. The measured Florida spread — 2 percent to 43 percent across eleven houses given the same treatment — is the honest picture.
Common belief
Cool roofing is a scam in the north because of the winter heating penalty.
What is actually true
The penalty is real and DOE says so, but it is smaller than the symmetric intuition suggests, because winter sun is lower and the days are shorter. BASC still finds cool roofs usually slightly beneficial in zones 5 and 6 for older air-conditioned homes, and only an overall penalty at zone 7 and above. “Not worth paying extra for” and “actively harmful” are different findings and they apply in different places.
Common belief
More attic ventilation will make my shingles last longer.
What is actually true
Ventilation is a minor factor in shingle temperature: 2 to 3 percent cooling effect in the Illinois measurements, against 20 to 30 percent for colour. Building Science Corporation puts colour and orientation at double to triple the durability effect of venting or not venting. Ventilation has other real jobs; this is not the strong one, and it is the one most often used to sell vents.
Common belief
A CRRC rating means the product met a standard.
What is actually true
The Council states the opposite in as many words: it does not require rated products to meet or exceed a minimum value, and it does not certify products, because certification implies a product meets or exceeds a standard. Any product can carry a rating regardless of how it performed. The rating is the number, not the verdict — you have to read the number.
Common belief
Cool roofs have to be white.
What is actually true
Most of the sun’s energy is not visible light, so a surface can be dark to the eye and still reflect strongly in the near infrared. BASC notes roofing products of any colour can be made to reflect near-infrared radiation, and gives dark metal finishes such as brown and red reaching three-year aged SRIs above 30. That is well short of white, and well ahead of a conventional dark surface.
Common belief
My roof gets to about the temperature the forecast says, so 100°F days are not that bad.
What is actually true
Air temperature is measured in shade in a ventilated screen. The EPA’s figure for dry exposed surfaces in full sun is 50 to 90 °F above the air. A dark panel on an insulated building is described in one manufacturer’s design manual as reaching 180 °F. The forecast is not the input to any of the ageing processes on this page.
Common belief
Heat damage will show up as something my insurer can be asked about.
What is actually true
There is no event, no date of loss, and no sudden change. Heat and UV ageing is the ordinary consumption of a covering’s life. Whether any specific policy responds to anything is governed by that policy and by the law of the state it was written in, and this page cannot tell you that — but the absence of an event is why this hazard is a purchasing decision and a maintenance decision rather than a claim.

How it actually fails

Granule loss, then accelerating embrittlement: The granules are the UV shield over the asphalt. As they shed, bare asphalt is exposed, UV and oxidation attack it faster, the surface becomes more brittle, and brittle surfaces release more granules. It is a feedback loop, not a linear decline, which is why the last few years of a shingle roof look much worse than the first twenty.What you can see: Slopes that have gone matte and slightly darker; a visible difference between the sunny and shaded elevations of the same roof; grit at the bottom of the downspout after rain.
Thermal splitting: Aged, embrittled material contracts on a cold night and can no longer accommodate the movement, so it cracks. On a membrane the split typically follows a deck joint underneath; on a shingle it runs across the unit.What you can see: Cracks that are conspicuously straighter than impact damage. A straight line is almost never something that fell on the roof.
Sealant strip failure on shingles: Asphalt shingles rely on a heat-activated sealant strip to bond each course to the one below. Heat is what seals it in the first place; heat over many years is also what ages the bond, and an aged bond releases in wind. This is the crossover point where the heat page becomes a wind page.What you can see: Tabs that lift in a moderate breeze; a slope that looks slightly ruffled from the pavement after a windy night.
Fastener-hole elongation on through-fastened metal: A through-fastened panel is pinned at every screw, so daily expansion and contraction has to go somewhere. One manufacturer’s design manual states that thermal contraction elongates the fastener holes by approximately 1/8 in. in extreme situations, and that a properly installed fastener with a sealing washer of adequate diameter is what stops that becoming a leak. The washer is therefore a consumable, not a permanent part.What you can see: Streaking below screw lines; visibly backed-out or over-driven screws; a re-torque or re-screw item appearing on a maintenance proposal, which is normal for this panel type and not a sign of a defect.
Gasket, boot and exposed-sealant ageing: Small polymer parts have small stabiliser packages and a lot of surface area relative to their mass. They reach the end of their UV life long before the field of the roof does.What you can see: Cracked collars around plumbing vents; hardened, shrunken sealant at counterflashings and reglets; interior stains near a penetration on a roof that is otherwise sound.
Blistering: Trapped moisture or gas expands as the surface heats and lifts the material above it. Intact blisters are cosmetic; opened ones have exposed the layer underneath and have become an entry point.What you can see: Raised bubbles, and dark open craters where they have burst.
Loss of reflectance: Soiling, biological growth and chalking of the surface reduce solar reflectance over time. Older field work found white surfaces could lose as much as 20 percent of their reflectance in two years, with degradation slowing after that.What you can see: A white roof that is grey, streaked, or patchy from an upstairs window or an aerial image. The energy penalty is invisible; the surface is not.

Sources and further readingSection link

Understanding Roofing / Published August 26, 2026

Scope and limitations

It cannot tell you what your roof surface reaches.
Surface temperature depends on orientation, slope, shading, colour, the assembly beneath, wind, and humidity.
The 150–190 °F band used on this page is the EPA's published surface-over-air range for dry exposed surfaces added to a 100 °F air temperature — an illustration of magnitude, not a measurement of any roof.
EPA's roofing-specific statement elsewhere, as much as 66 °F above the surrounding air, is narrower than the 50–90 °F range that band is built from.
It cannot tell you what a cool roof would save you.
The published percentages differ by a factor of three depending on climate zone, ceiling insulation, duct location, and roof type, and the measured Florida spread across eleven identically treated houses was 2% to 43%.
It does not publish an installed cost for any roof, cool or conventional.
The only price statement here is DOE's, that cool asphalt shingles cost about the same as conventional ones.
It does not publish a service-life number for any covering in any climate.
Heat and UV move a covering within a planning range; where it lands depends on the product, the assembly, the orientation, and the maintenance, and a single number without those attached would be decoration.
It cannot tell you what your energy code requires.
The California figures quoted are from a superseded edition and are shown only to illustrate how such a rule is structured.
Every code claim is a jurisdiction question.
It does not model your building.
The Phoenix and Seattle comparison holds colour, slope, orientation, and assembly constant, and no two real houses do that.
It cannot tell you whether any insurance policy responds to anything.
This page contains no coverage guidance because heat and UV ageing produces no loss event to make a claim about.

Consumer Guide to Cool Roofs (DOE/EE-2478)
U.S. Department of Energy, Energy Saver / August 2021
That solar reflectance is the amount of sunlight reflected and thermal emittance the efficiency with which a surface cools itself by letting go of absorbed heat, both measured 0 to 1, the higher the cooler; that a white or cool-colour roof reflects up to 90% of the sunlight that hits it; that substituting a cool roof for a conventional roof in warm or hot climates can reduce annual air conditioning energy use of a single-storey building by up to 15%; that installing a cool-colour roof on a home can shave at least 5 to 20% from cooling costs, greatest in hot and warm climates; that cool asphalt shingles cost about the same as conventional asphalt shingles; that it is rarely economical to replace a mechanically sound roof just to increase its solar reflectance; that a cool roof may increase the need for heating energy in winter, and that even in mixed climates winter penalties can be smaller than summer savings; that shingles lose some ability to reflect over time so savings should be based on long-term values; that the three cost-effectiveness factors are aged performance, incremental initial cost, and the cost of keeping the roof clean; and the product split at a pitch of 2:12 between low-slope and steep-slope cool options.
A consumer guidance leaflet dated 2021. It gives no installed prices, no climate-zone breakdown, and no method behind its percentage ranges; it is a statement of direction and order of magnitude, not a calculation for any building.
Cool Roofs and Walls to Reduce Heat Gain
U.S. Department of Energy, Building America Solution Center (PNNL)
The definition of a cool roof or wall; that a standard asphalt shingle roof on a Florida home heats to 140°F; the definitions of solar reflectance, thermal emittance and SRI, and that SRI typically ranges 0 to 100 with higher values indicating a cooler surface; that SRI can be reported as initial or three-year aged values and that aged values based on three years of natural weathering are the industry standard and what many codes and standards require for compliance; the three-year aged SRI ranges — as high as 85 or 90 for white metal finishes, above 30 for dark metal colours such as brown and red, generally about 20 to 35 for cool asphalt shingles, well into the 90s for white membrane finishes and about 5 to 20 for dark membranes; that roofing products of any colour can be made to reflect near-infrared radiation; that cool roofs reduce cooling energy but increase heating energy, with the increase generally much less than one might think because solar impact on roof and wall temperature is lower in winter than summer; and the climate-zone summary — almost always beneficial in zones 1A through 3B, usually beneficial but reduced in 3C, 4A and 4B, nearly zero savings over a year in 4C, usually slightly beneficial for older air-conditioned homes in zones 5 and 6, and generally an overall energy penalty in zones 7 and above.
Best-practice guidance for builders and designers, not adopted law and not a code determination anywhere. Its climate-zone conclusions are general findings for typical construction, not a prediction for a specific building.
Cool Roof (building science measure)
U.S. Department of Energy, Building America Solution Center (PNNL)
That cool roofing maintains lower attic temperatures by reflecting sunlight; that standard or dark roofs can reach temperatures of 150°F or higher in hot climates and that a cool roof under the same conditions could stay more than 30°F cooler; and that cool roofing materials certified under the ENERGY STAR programme include metal, tile and asphalt shingles meeting minimum criteria for solar reflectance.
A short measure summary. It gives no climate-zone conditions, no heating-penalty statement, and no code citation.
Using Cool Roofs to Reduce Heat Islands
U.S. Environmental Protection Agency
That a high solar reflectance, or albedo, is the most important characteristic in how well a cool roof reflects heat away from a building, and that a high thermal emittance also plays a role, particularly in climates that are warm and sunny; that cool roofs can lower maximum indoor temperatures in non-air-conditioned residential buildings by 2.2 to 5.9°F and reduce peak cooling demand in air-conditioned residential buildings by 11 to 27%; that cool roofs may increase energy use in winter in cold climates but that this heating penalty is typically offset by summer cooling savings, mitigated by lower winter sun angles, shorter days, effective insulation and snow cover; and the split of cool product types between low-slope and steep-slope roofs.
Written for the urban heat island programme rather than for a homeowner choosing a roof. Its indoor-temperature and peak-demand figures come from cited studies of particular building stocks and are not transferable to a specific house.
Reducing Urban Heat Islands: Compendium of Strategies — Urban Heat Island Basics
U.S. Environmental Protection Agency
That on a hot, sunny summer day the sun can heat dry, exposed urban surfaces, like roofs and pavement, to temperatures 50 to 90°F (27 to 50°C) hotter than the air, while shaded or moist surfaces remain close to air temperatures — the figure this page uses to derive its illustrative surface-temperature band.
The PDF posted at this address carries a “DRAFT” running header on its pages, as does EPA's 2017 reposting of the same chapter; the 50 to 90°F figure appears twice within the document. EPA's current public heat-island pages do not repeat it — they give a narrower roofing-specific figure instead, which is why this page prints both. It is a statement about dry exposed surfaces generally, not a measurement of any roof assembly, and it does not distinguish roofing materials.
What Are Heat Islands?
U.S. Environmental Protection Agency
That on a warm day conventional roofing materials may reach as much as 66°F warmer than the surrounding air temperatures — EPA's roofing-specific figure, printed on this page alongside the broader 50 to 90°F range for dry exposed surfaces so the reader can see the difference between them.
A short public explainer page. It gives no assembly, no colour, no location and no measurement method behind the 66°F figure, and it is not a prediction for any roof.
Heat — Heat-Related Illnesses and First Aid
U.S. Department of Labor, Occupational Safety and Health Administration
The heat-stroke signs and first-aid steps in the safety callout: confusion, slurred speech, unconsciousness, seizures, heavy sweating or hot dry skin, very high body temperature and rapid heart rate; that when those symptoms are present you call 911 immediately and cool the person with ice or cold water until help arrives; that you move them to a cooler area such as shade or air conditioning; and that you never leave someone with a heat-related illness alone because the illness can rapidly become worse.
Occupational guidance written for employers and workers, not medical advice and not a diagnosis. It does not replace emergency medical services, and this page is not a substitute for a heat illness prevention plan.
Measured and Simulated Performance of Reflective Roofing Systems in Residential Buildings (FSEC-PF-331-98)
Parker, Huang, Konopacki, Gartland, Sherwin and Gu — Florida Solar Energy Center and Lawrence Berkeley National Laboratory, in ASHRAE Transactions Vol. 104 Pt. 1 / 1998
That in tests on 11 Florida homes with the roof colour changed mid-summer the average measured cooling energy use was reduced by 19%, ranging from a low of 2% to a high of 43%; that laboratory testing of 14 different colours of asphalt shingles found solar reflectances varying only from 5 to 25%, that six of the more popular grey and brown colours had a tested average solar reflectance of only 10%, and that a specially prepared shingle manufactured to be more white had a tested reflectance of only 31%, because of the dark asphalt substrate on which the granules are impregnated; that several studies indicate white surfaces can lose as much as 20% of their reflectance in two years with further degradation slowing after that point; the Miami modelling results — 26% cooling savings with no ceiling insulation, 14% (730 kWh) with R-19 and an attic duct system, and 8% (362 kWh) with R-19 and ducts within the conditioned space; that heating requirements in existing housing were increased by 3% in Miami to 6% in San Francisco; that except in the northernmost locations (Minneapolis and Detroit) and cool cloudy locations (Seattle and San Francisco) the combined cost of heating and cooling was lower for reflective roof surfaces; the Detroit existing-construction annual space-conditioning costs of $0.508/ft² for the base case against $0.504/ft² for a new white roof; that the locations where a new reflective roof produced greater annual energy cost savings than adding another increment of insulation — Los Angeles, Atlanta, Houston, Ft. Worth, Miami, Fresno and Phoenix — are all below 37 degrees north latitude; and that in climates where heating needs dominate, above 40 degrees north latitude, there was little advantage from reflective roofing and added insulation was clearly a better choice in the coldest locations.
Published in 1998. Its dollar figures use 1996 national average energy prices ($0.084/kWh and $0.63/therm) and its prototype buildings use equipment efficiencies and insulation levels of that era, so the absolute costs are obsolete. Measurements were made in Florida only; the other 13 locations are DOE-2.1E simulation. Its aged-reflectance assumptions are explicitly described by the authors as a simple conservative assumption in the face of uncertainty.
Venting of Attics and Cathedral Ceilings
William B. Rose (University of Illinois) and Anton TenWolde (USDA Forest Service, Forest Products Laboratory), ASHRAE Journal, October 2002 / October 2002
That most asphalt shingle manufacturers warrant their products only for ventilated roofs, and that such warranty clauses date from the 1980s; that the published rationale holds that venting cools shingles and thereby affects the rate of embrittlement by reducing the rates of oxidation and volatilisation of asphalt hydrocarbons; that ventilation is a minor factor in the determination of shingle temperature; that ventilation of a black-shingle, truss-framed roof assembly in Illinois has a 2% to 3% cooling effect on shingles while the effect of colour is 20% to 30%; that Simpson and McPherson found white roofs were up to 36°F cooler than grey roofs and up to 54°F cooler than brown roofs; that Cash and Lyon showed through computer modelling that shingle temperature is more strongly a function of geographic location, the direction a roof faces, and surface colour than ventilation, and that venting cools shingles but the cooling effect is not strong; and that roof vents may play a detrimental role in forest fires in residential areas by providing entry for sparks and burning brands.
A 2002 review article assembling other researchers' findings; the shingle-temperature measurements it reports are from Illinois and from studies of particular assemblies. It is not adopted code anywhere, and it does not evaluate current cool-shingle products.
BSD-102: Understanding Attic Ventilation
Joseph Lstiburek, Building Science Corporation / 25 October 2006
That shingles installed on unvented attic assemblies operate at a slightly higher temperature — a 2 or 3 degree F rise in average temperature for asphalt shingles and a corresponding 10 degree F rise in average temperature for sheathing, citing Parker and Sherwin 1998, Rudd and Lstiburek 1998 and TenWolde and Rose 1999; that applying the Arrhenius equation a 10 percent reduction in useful service life should be expected, comparable to the effect of the installation of radiant barriers; and that the colour of shingles and roof orientation have a more profound effect on the durability of shingles than the choice of venting or not venting — double or triple the effect.
A building-science digest, not a code or a test report. The 10 percent service-life figure is the author's application of the Arrhenius equation to a small average temperature difference, presented as an expectation rather than a measured outcome, and this page presents it that way.
A Simple Guide to Understanding Your Asphalt Shingle Roofing System
Asphalt Roofing Manufacturers Association (ARMA)
That granules are what provide the colouring to shingles and that they also protect the shingle's asphalt from degradation that would occur from prolonged exposure to UV light; that the primary purpose of the asphalt is to serve as a water-shedding agent and to hold the granules in place; and that proper ventilation consisting of balanced intake and exhaust helps extend the service life of an asphalt shingle roof by reducing attic temperature and removing condensation.
Consumer-facing guidance from a manufacturers' trade association. It is not a test method, not a standard, and not a code, and its statement about ventilation extending shingle life is not quantified — the research sources above put a small number on that effect.
Roof Rating Program FAQs
Cool Roof Rating Council
That the CRRC does not require rated products to meet or exceed a minimum value for solar reflectance (SR) and thermal emittance (TE); that the CRRC does not certify products through the Roof Rating Program, as certification implies that a product meets or exceeds a standard; that products are given initial testing followed by aged testing, both by a CRRC-approved test lab; and that minimum SR and TE requirements vary among jurisdictions.
The rating body describing its own programme. It establishes what a rating is and is not; it does not establish what any jurisdiction requires, and it does not evaluate durability, watertightness, or energy savings.
New CRRC Rapid Ratings Process Simulates Three-Year Weathering in Mere Days
Cool Roof Rating Council / 11 August 2015
That all CRRC rated products must still complete three-year weathering, which is critical for informing consumers of product performance over time; that ASTM D7897 is the Standard Practice for Laboratory Soiling and Weathering of Roofing Materials to Simulate Effects of Natural Exposure on Solar Reflectance and Thermal Emittance; and that Rapid Ratings are displayed until a product completes the required three-year field weathering process, at which point they are replaced by the field-exposed aged values.
A programme announcement from 2015. Programme details may have changed since; the CRRC-1 Roof Product Rating Program Manual is the controlling document.
California Building Energy Efficiency Standards (Title 24, Part 6) — cool roof summary
Cool Roof Rating Council
That the current edition is the 2025 Building Energy Efficiency Standards for Residential and Nonresidential Buildings (Title 24, Part 6) with an effective date of 1 January 2026; that the prescriptive residential requirements sit in Sections 150.1(c)11 and 150.2(b)1I; and that any roofing product used for compliance must be rated by the CRRC.
A summary of law written by the rating body whose labels are used to demonstrate compliance with it. It is used here only to locate the relevant sections and to corroborate the edition and effective date already confirmed on the California Energy Commission's own pages. It is not adopted law, and no numeric threshold on this page is taken from it.
2025 Building Energy Efficiency Standards
California Energy Commission
That buildings whose permit applications are applied for on or after 1 January 2026 must comply with the 2025 Energy Code — the reason the 2022 thresholds quoted on this page are described as superseded.
The adopting authority's landing page for the edition. It is authoritative for the edition and effective date and not a substitute for the adopted text, which contains the requirements themselves. It governs California only, and local jurisdictions may adopt ordinances that go beyond it.
2022 Building Energy Efficiency Standards
California Energy Commission
That the 2022 Energy Code (Title 24, Parts 1 and 6) applied to buildings whose permit applications were applied for on or after 1 January 2023 — establishing the edition and window that the permit guide below describes.
Superseded for permit applications made on or after 1 January 2026. Cited here to date the edition, not to state a current requirement.
Residential Roofing: Guide to 2022 Energy Code Requirements
BayREN (Bay Area Regional Energy Network) Codes & Standards Program, as published by the City of Benicia, California / Last updated 1 May 2023
That under the 2022 edition cool roofing requirements were triggered when 50% or more of the roof was being replaced; that compliance was demonstrated with a CRRC product label; and the exemption list for steep-slope roof replacements — an attic radiant barrier meeting Section 150.1(c)2, no ducts in the attic in climate zones 2, 4, 9, 10, 12 and 14, a ceiling insulated to at least R-38 or a ceiling assembly U-factor of 0.025 or lower, or at least R-2 continuous insulation above or below the roof deck — which is the point this page draws from it.
A regional public-agency permit guide, not the adopted text of Title 24 Part 6, and it describes the 2022 edition which is superseded for permits applied for on or after 1 January 2026. Written for Bay Area jurisdictions. Nothing in it is a requirement anywhere outside California, and it is quoted on this page to show how such a rule is structured rather than to state a current threshold.
Structural Standing Seam Roof Systems Design Manual (SRS), © 2011
CENTRIA / 2011
The coefficients of expansion given in inches per inch per 100 degrees Fahrenheit — carbon and galvanised steel .00065, stainless steel .00099, aluminium .00128; that an unpainted or light-coloured panel on an uninsulated building exhibits the least thermal movement and that a minimum 100 degree Fahrenheit temperature differential should therefore be figured; that the greatest movement occurs with a dark-coloured panel on an insulated building and that the exposed metal temperature in this condition can reach 180 degrees Fahrenheit; that thermal contraction on through-fastened roofing will elongate fastener holes in the panels by approximately 1/8 in. in extreme situations, with a properly installed fastener and sealing washer preventing leakage at those points; that linear thermal panel movement is cumulative and the force of that movement is sufficient to buckle and tear metal and shear fasteners if long-length panel movement is restricted; and that a panel-and-clip assembly cycled 100,000 times, simulating up to five cycles per day for 50 years, showed a maximum total material loss at the clip contact points of 0.005 in.
One manufacturer's design manual for its own proprietary system, dated 2011 and product-specific throughout. Its coefficients are standard values for the metals named, but its temperature assumptions, cycle test, and fastener-hole figure describe its own products and test programme, not roofing generally. It is not a code, not a standard, and not a design determination for any building.
1991–2020 U.S. Climate Normals, monthly — Phoenix Sky Harbor International Airport, AZ (USW00023183)
NOAA National Centers for Environmental Information / 1991–2020 normals period
The Phoenix monthly normal daily maximum and minimum temperatures used in the two-cities comparison, including a July normal daily maximum of 106.5°F, a July normal daily minimum of 84.5°F, a December normal daily minimum of 45.3°F, and normal daily maxima at or above 90°F in May, June, July, August and September.
Normals are 30-year averages of daily values at one airport station. They are not extremes — record highs are considerably higher — and they do not describe any neighbourhood, any microclimate, or any roof.
1991–2020 U.S. Climate Normals, monthly — Seattle–Tacoma International Airport, WA (USW00024233)
NOAA National Centers for Environmental Information / 1991–2020 normals period
The Seattle monthly normal daily maximum and minimum temperatures used in the two-cities comparison, including an August normal daily maximum of 77.6°F, an August normal daily minimum of 57.2°F, a December normal daily minimum of 37.1°F, and no month whose normal daily maximum reaches 90°F.
As above. A single airport station, thirty-year averages of daily values, and no information about cloud cover, solar irradiance, or the hours of direct sun a particular roof plane receives — all of which matter to the surface temperatures this page discusses.
Condensation Risk of Mechanically Attached Roof Systems in Cold-Climate Zones
Manfred Kehrer and Simon Pallin, Oak Ridge National Laboratory, presented at the 28th RCI (now IIBEC) International Convention and Trade Show, March 2013 / 2013
That a white exterior surface or so-called cool roof, applied to decrease cooling loads and save energy, increases the condensation risk due to lower energy gains during daylight hours; that flat roofs generally have a high potential of nightly overcooling and therefore an increased risk of condensation within the construction, particularly in cold-climate zones; and that across 128 simulated scenarios the amount of moisture accumulated below the membrane was almost doubled in a cool roof construction compared with a traditional black roof.
A conference paper on mechanically attached low-slope membrane systems specifically. It does not describe steep-slope residential assemblies, and its findings are the output of hygrothermal simulation for defined constructions and climates rather than a general rule about reflective surfaces. Its four climates span zones 4 to 7 only, and within that span the authors found climate the least influential of their four variables — indoor moisture supply and air intrusion rate mattered more, so this is a whole-assembly question rather than a surface-colour one.

Suggest a correction