“When do roofs need replacing?” has at least five honest answers, and the reason the internet gives one — a 15-to-30-year range, unsourced, on every contractor blog — is that it does not say who is asking. A warranty answers it one way, a permit record another, a state a third, a material a fourth, and the house itself a fifth. They disagree with each other, and the disagreement is the useful part, because it is what tells a roofing team where to look.
This page gives the five answers with their sources, explains why none of them can be turned into a prediction about a particular roof, and then does the thing the five answers are actually good for: finding the cohort of homes where replacement is most likely in the next couple of years. It is written for roofing reps and the people who build their territories.
Disclosure: Scout Data is our product, and the permit database the second and third answers come from is ours — the same records our homeowner files and Scout field app put on each home. The answers hold whatever tool you use to act on them.
Answer one: the warranty’s
Shingle warranties are written in decades, and they are the number most homeowners have in their heads. They are also a product claim under specified conditions — installed to the manufacturer’s instructions, ventilated as the instructions require, on a roof that has not been hit by anything — and most are prorated after the first years. A warranty term is the upper bound of what a roof might do. It is not what roofs, on average, actually do.
Answer two: the permit record’s — 14 years, and a floor
Where the same home shows two roofing permits, the gap between them is a direct observation of a replacement interval, which is something no survey measures. Across 309,255 such intervals of three years or longer in Scout Data’s permit database, the median is 14 years, with quartiles at 7 and 20. The full derivation is in roof replacement statistics.
Two things about that number matter more than the number. First, the quartiles: a quarter of observed replacements happen inside seven years and a quarter after twenty, so “14” describes a wide band, not a point. Second, the censoring: a jurisdiction that started publishing permits in 2008 cannot show a 25-year interval, so long-lived roofs are under-represented and the median is a floor on true roof life, not an average of it. The warranty and the permit record are not contradicting each other. One is a ceiling and the other a floor.
Answer three: the state’s — 8 to 16
The national median hides the number that actually decides a territory. Median observed interval between consecutive roofing permits on the same home, by state:
| State | Median observed interval |
|---|---|
| Ohio | 8 years |
| Texas | 9 years |
| Colorado | 10 years |
| Michigan | 11 years |
| Minnesota | 12 years |
| California | 12 years |
| Florida | 16 years |
Twice as long in Florida as in Ohio, for the same product. The pattern is a weather-and-insurance map. Ohio, Texas and Colorado sit in or near the hail belt, where a storm claim funds a replacement long before the shingles wear out, and each storm resets the clock. Florida is a hurricane-code market with more tile and metal on its roofs and a replacement cycle shaped by insurers and building code rather than by wear.
Florida also shows how an insurer can put a number on the calendar directly. Under Florida Statutes §627.7011(5), as enacted in 2022, an insurer may not refuse to issue or renew a homeowner’s policy solely because of roof age if the roof is under 15 years old; for a roof 15 or older, the homeowner may obtain an inspection, and a finding of five or more years of remaining life has the same protection. That is a rule about coverage, not about roofs — but it means a large number of Florida roofs have a conversation about their age scheduled for their fifteenth year whether or not they need one. Check the current text before quoting it; statutes move.
Answer four: the material’s
Roof material is recorded by many assessors, unlike roof age, and it changes the schedule. Across 50,687,261 single-family homes with a recorded material, 77.4% carry asphalt or composition shingle, 6.3% metal, 5.4% tile or concrete, 3.2% wood shake and 0.4% slate. For most of the country that means the interval figures above are asphalt figures. Where the exceptions cluster they overturn the schedule entirely: Nevada is 69.7% tile or concrete, Vermont 29.3% metal, and a tile or standing-seam roof on a twenty-year-old house is not due on an asphalt timetable. Roofing material statistics has the state and decade breakdown.
Answer five: the house’s — mostly useless, with one exception
The median home is 42 years old when a roofing permit is pulled on it, and the median U.S. single-family home was built in 1979 — 47 years old, per American housing stock age. Against a 14-year interval, the typical house is on its third roof, and its build year tells you nothing about when that roof went on. Build year is the weakest of the five answers and the one cheap “roof age” fields are usually made of.
The exception is young stock. A home under about 25 years old is plausibly on its original roof — and a quarter of roofing permits go to homes under 25. On a subdivision built between, say, 2004 and 2007, every roof went on in the same three years, and the block will come due as a block. That is the one case where the house’s answer is worth something, and it is worth a lot.
Why none of this predicts a roof
Put the five answers together and the temptation is to score each home with a “years remaining” figure. Resist it. The interval band is seven to twenty years around the median; a hailstorm can move a roof from “fifteen years left” to “replaced this spring” in an afternoon; an unpermitted replacement leaves no record at all; and a south-facing slope in Arizona ages differently from a shaded one in Maine on the same house. Every one of those is invisible to the record.
What the record supports is a statement about cohorts: homes in this state, past this permit age, on this material, in this build-year band, have a high base rate of replacement in the next few years. A rep who says “our records show your roof is due” has turned that into a claim about one roof, and the homeowner whose roof was quietly replaced in 2021 will correct them at the door. The honest version — “the last roofing permit we can see is from 2009; is that right?” — is also the one that books.
How to target it
Each of the five answers becomes one filter, applied in order.
- Hold the threshold per state, not per product. Ten years since the last permit is past the median in Ohio and Texas and well short of it in Florida. A single national cutoff over-targets one and under-targets the other.
- Use the permit date where there is one, and suppress the recent ones. In hail markets the last permit is often a storm date, and a home that had a claim-funded replacement three years ago is a stop, not a lead. Where there is no permit, how to find homes with old roofs covers how far to trust condition and build year instead.
- Filter on material before applying an asphalt threshold. Run the schedule across Nevada without it and the territory is tile.
- Work young stock as blocks. Where a subdivision’s build years are tight and the first roofs are reaching the local median, the cohort is the street, and the first permit on it is the signal that the rest are coming.
- Add a trigger. Age puts a home in the cohort; a sale, a purchase, a listing, a storm passing over, or — in Florida — a roof approaching its fifteenth year gives the homeowner a reason to act this year rather than eventually.
How the resulting list is worked depends on which trigger fired. Where the trigger is the calendar, how to sell a roof without a storm covers the retail motion — the inspection as diagnosis, pricing with the homeowner’s money, the follow-up cadence. Where the trigger is a storm, the same cohort filtered to a footprint is the list in storm damage leads, and the fact that the permit date separates the roofs that are due from the ones replaced after the last storm is why the two motions share a database and not a script.
Frequently asked questions
How often do roofs need replacing?
There is no single number, and anyone who gives you one without a method is guessing. The closest thing to a measurement: among U.S. homes where two or more roofing permits are observed, the median gap between them is 14 years, across 309,255 observed intervals — a quarter at 7 years or shorter, a quarter at 20 or longer. That median is a floor on real roof life, because long-lived roofs rarely show both permits inside a records window.
Why do roofs in some states get replaced so much sooner?
Weather and insurance, not shingle chemistry. Hail-belt states replace early because insurance-funded storm claims reset the clock — Ohio, Texas and Colorado run 8-to-10-year median intervals. Florida runs 16, a hurricane-code market heavy on tile and metal with replacement cycles shaped by insurers and building code. The same asphalt shingle has a different expected life in each of those places.
Can data predict when a specific roof will need replacing?
No. The interval distribution is wide — a quarter of observed replacements happen inside 7 years and a quarter after 20 — and any one roof is somewhere in that band for reasons the record cannot see: the storm that hit it, the slope that faces south, the work done without a permit. What data can do is find the cohort in which the base rate is high. That is a targeting statement, not a prediction, and the page says so because a rep should too.
Does the age of the house tell you the age of the roof?
Almost never, past the first roof. The median U.S. single-family home was built in 1979 and is 47 years old; against a 14-year replacement median, the typical house is on its third roof and its build year says nothing about when that roof went on. The exception is young stock: a home under about 25 years old is plausibly on its original roof, and a quarter of roofing permits go to homes under 25.