Two things make a roof more likely to be a problem after a storm: stones big enough to damage it, and a house old enough that the roof, decking and flashing have had time to wear. We matched both for 542 US counties with at least 20,000 housing units, using NOAA golf-ball hail days (1.75 inches or larger, 2014 to June 2026) and Census data on when homes were built. The result is not what we expected: the two risks barely overlap.
We set two lines. A county has an old housing stock if at least 60% of its homes were built before 1980; 129 of the 542 counties (23.8%) qualify. It has heavy big-hail exposure if it logged 25 or more golf-ball days; 18 counties (3.3%) qualify. Only 1 county does both: Potter County, TX (Amarillo), with 49,133 housing units, 70.8% of them built before 1980, and 27 golf-ball days.
The picture, county by county
The cloud has no slope to speak of. The correlation between the share of pre-1980 homes and golf-ball days across these counties is -0.11, which is close to zero. Hail counties lean newer: the median county in the set has 47.3% of its homes built before 1980 and 5 golf-ball days. Several of the hardest-hit counties built most of their housing after 1980: Denton County, Texas has 12.8% pre-1980 homes and El Paso County, Colorado 36.2%.
Where the big hail is
| County | Housing units | Days 1.75 in+ | Largest stone | Homes before 1980 | Median year built |
|---|---|---|---|---|---|
| Pennington County, SD | 50,655 | 83 | 4.50 in | 48.1% | 1982 |
| El Paso County, CO | 296,753 | 53 | 4.00 in | 36.2% | 1988 |
| Weld County, CO | 125,033 | 38 | 4.25 in | 30.8% | 2000 |
| Laramie County, WY | 45,032 | 37 | 4.25 in | 50.0% | 1980 |
| Denton County, TX | 364,120 | 31 | 5.90 in | 12.8% | 2002 |
| Lea County, NM | 28,125 | 30 | 3.00 in | 55.0% | 1977 |
| Sedgwick County, KS | 225,795 | 29 | 2.75 in | 53.8% | 1977 |
| Lubbock County, TX | 136,468 | 28 | 4.50 in | 45.1% | 1984 |
| Tarrant County, TX | 827,874 | 27 | 5.00 in | 33.8% | 1989 |
| Randall County, TX | 61,137 | 27 | 3.00 in | 42.4% | 1986 |
None of these ten counties reaches 60% pre-1980 housing; Lea County, New Mexico comes closest at 55.0%. The stones are real: the largest hail in any of the 542 counties is 6.60 inches (Kankakee County, IL). The median build year in these ten counties runs from 1977 to 2002.
Where the old houses are
| County | Housing units | Homes before 1980 | Median year built | Days 1.75 in+ |
|---|---|---|---|---|
| St. Louis city, MO | 173,920 | 85.9% | 1938 | 5 |
| Knox County, IL | 23,788 | 82.2% | 1956 | 4 |
| Cuyahoga County, OH | 615,513 | 81.0% | 1957 | 3 |
| Milwaukee County, WI | 423,838 | 79.8% | 1956 | 3 |
| Delaware County, PA | 229,390 | 79.7% | 1957 | 0 |
| Henry County, IL | 22,208 | 79.2% | 1961 | 7 |
| Broome County, NY | 92,889 | 78.2% | 1959 | 5 |
| Beaver County, PA | 79,159 | 78.0% | 1958 | 0 |
| Whiteside County, IL | 25,850 | 77.7% | 1960 | 4 |
| Allegheny County, PA | 604,558 | 77.2% | 1957 | 4 |
The ten oldest counties here average 3.5 golf-ball days, against a set-wide median of 5. These are Missouri, Illinois, Ohio, Wisconsin, Pennsylvania and New York counties, where golf-ball hail is infrequent. For the oldest-housing counties, our data say hail is not the dominant exposure; what else wears their roofs is outside this dataset.
Do not read a high hail count as "my old roof is doomed" or a low count as "my old roof is fine". The data say where risks sit, not what your roof has been through. Housing age counts the building, not the roof; a 1962 house may have its third roof, and a 2004 house may still have its original. The fastest way to know is to ask the seller or previous owner for the reroof date, check for permits, and have the roof looked at after any reported storm. The article on roof age and hail covers the repair-or-replace decision as a roof ages.
Near misses worth knowing
Loosen the lines to 50% older homes and 15 hail days and 25 counties appear, among them Laramie County, WY (50% pre-1980, 37 days), Lea County, NM (55% pre-1980, 30 days), Sedgwick County, KS (54% pre-1980, 29 days), Potter County, TX (71% pre-1980, 27 days), Comanche County, OK (56% pre-1980, 27 days). If your house is in one of them, both risks are material there, if less extreme than Amarillo's combination. For a county not listed here, look up its own record on a county hail history page.
Frequently Asked Questions
Does an older house mean more hail damage?
Not by itself. Across 542 large counties, the share of pre-1980 homes and the number of golf-ball hail days are almost uncorrelated (-0.11). Damage depends on the specific roof, which the Census does not measure.
Which large county has both heavy hail and old housing?
In this set only Potter County, TX, which passes both thresholds (70.8% of homes before 1980, 27 golf-ball days).
Why only counties with 20,000 housing units?
Small counties have noisy percentages and few reports, so one storm or one new subdivision can change the picture. The cutoff keeps the comparison to counties with enough homes to say something.
Is "built before 1980" the same as having an old roof?
No. The Census counts when the building was built. Roofs are replaced, often more than once, so the age of the structure is only a rough pointer.
Where the numbers come from
- NOAA NCEI Storm Events Database
- U.S. Census Bureau, ACS table B25034 (year structure built)
- U.S. Census Bureau, ACS table B25035 (median year built)
Data: NOAA hail reports by county, 2014 to June 2026; Census ACS 5-year estimates, 2019 to 2023, county level; counties with at least 20,000 housing units. Processed by PickARoofer. Method and limits: how we build our storm and hail figures.
Sources
- NOAA NCEI Storm Events Database (ncei.noaa.gov)
- U.S. Census Bureau, ACS table B25034 (year structure built) (data.census.gov)
- U.S. Census Bureau, ACS table B25035 (median year built) (data.census.gov)
How we write and fund our articles: editorial policy.
Ready to Get Contractor Quotes?
Compare local roofing contractors in your area. Free, no-obligation estimates from licensed professionals.