Before Regret · Research
More than a quarter of every flood insurance claim the United States has ever paid was on a property whose policy was rated outside the high-risk flood zone. That is the zone each policy was rated in at the time of the claim, across records going back to the 1970s — not a statement about where any property sits on today’s map.
Interactive · 1,921 counties
Every county with at least 25 classifiable NFIP claims on record. Type a county name to see what share of its flood claims were paid outside the mapped zone.
Share of flood claims paid outside the mapped high-risk zone
A county's percentage is the share of its own claims history that fell outside the mapped zone. It is not a forecast, and a county with few claims on record will swing widely from one storm to the next.
A standard homeowners policy does not cover flood. To be covered you need a separate policy, and the practical trigger for buying one is a line on a FEMA map: inside a Special Flood Hazard Area, a federally backed mortgage requires it. Outside that line, almost nobody is required to buy, and almost nobody does.
Our earlier analysis of flood coverage, Risk Without Cover, found that in the typical county fewer than one home in six inside a mapped zone carries a policy, and noted a limit we could not then measure: flood maps are imperfect, and a large share of flood damage happens outside them. This is that measurement. The study before it, Risk Without Price, established the figure this one starts from: 59.2% of the building loss FEMA expects American homes to suffer each year comes from flooding, none of which a standard homeowners policy covers.
FEMA publishes every claim the National Flood Insurance Program has ever paid, 2,721,780 of them, each carrying the flood zone the policy was rated in. Counting them by zone answers a question the take-up figures cannot: when the NFIP actually pays, where was the property standing?
Of 2,578,413 claims that carry a usable zone code, 686,045 — 26.6% — were rated in zones X, B or C: the moderate-to-minimal risk designations, where flood insurance is not federally required. Those claims account for $20,627,923,968, or 23.4% of everything the programme has paid out.
The dollar share sits a little below the claim share, which is what you would expect: damage outside the zone tends to be shallower than damage in a coastal surge. It is not, however, small. $20.6 billion has been paid on properties whose policies were rated outside the high-risk zone.
The line on the map does not decide where water goes. It decides who was told to buy insurance.
The national figure hides enormous variation. Across the 47 states with at least 1,000 classifiable claims, the out-of-zone share runs from 10.2% in NJ to 54.8% in SD. The states at the top are mostly inland: SD 54.8%, TX 50.7%, CO 50.4%, ND 49.7%, MT 46.1%.
Texas is the case that matters most, because its sample is large enough that the number cannot be a fluke: 50.7% of 380,408 Texas claims were paid outside the mapped zone. In Harris County alone — Houston, and Hurricane Harvey — the figure is 54.8% of 167,261 claims. Independent assessments after Harvey reached the same conclusion by a different route: a large share of the homes that flooded were never in the floodplain on paper.
Here is what makes the first finding matter. Inside the mapped zone, where a mortgage usually compels it, 47.9% of homes carry flood cover. Outside the zone, where nothing compels it, the figure is 1.25% — 1,321,764 policies spread across 105,487,685 homes.
So 26.6% of all claims, and $20.6 billion, were generated by a group of properties of which only about one in eighty is insured at all. Every uninsured home outside the zone that flooded is simply absent from this dataset: no policy, no claim, no record.
Every figure here counts only people who bought a policy. Because out-of-zone take-up is 1.25%, the out-of-zone share of actual flood damage is certainly higher than its 26.6% share of flood claims.
We deliberately do not put a number on how much higher, and neither should anyone quoting this. The households that buy flood insurance when nobody requires them to are not a random sample of their neighbours — they are disproportionately people who have flooded before, or who know something about their ground that the map does not. That selection cuts against a simple scaling-up, and there is no way to size it from this data.
Read annually, this number is unstable, and anyone using it should know that. In 2017, the year Harvey put much of Houston under water, 42.3% of claims nationally fell outside the zone. In 2024, dominated by coastal hurricane damage in Florida, it was 10.1%. A coastal surge lands squarely in mapped zones; inland and flash flooding does not.
That is why the headline figure here is the long-run one across every claim on record rather than a recent year. A single year measures which storms happened, not how well the maps describe where American homes flood.
The zone is the rated zone, not today's map. This is the most important caveat on the page. ratedFloodZone records the zone a policy was rated in at the time, and these claims span the 1970s to 2026 — a period over which flood maps have been redrawn many times. A property rated X in 1994 may sit inside a mapped zone today, and the reverse also happens. This measures how well the map in force described risk when each claim was paid, which is the question a buyer faces, but it is not a statement about current maps.
NFIP only. Private flood insurance is not in this data. The private market has grown, so total flood claims are higher than the totals here by an amount this file cannot measure.
Nothing here explains why. Map vintage, terrain, drainage and development, rainfall patterns, and how strictly the purchase requirement is enforced all plausibly contribute. This is a count, not a causal account.
Counts are claims, not properties. A property that flooded three times contributes three claims. Repetitive-loss properties are therefore weighted more heavily than single-loss ones.
Some counties show no take-up figure. The coverage file counts residential structures inside mapped zones, and in 22 counties that count is too low to be right: taken at face value it implies more policies in force than there are homes to hold them. Collier, Florida reports 38,559 contracts against 152 counted structures. Those are structure-count errors rather than real coverage above 100%, so the take-up figure is suppressed for those counties in the lookup above. Their out-of-zone claim share is unaffected and still shown: it is computed from the claims file and does not use that denominator.
Ranked by the share of claims paid outside the mapped zone. Dollar share is the same calculation applied to amounts paid rather than claim counts.
| State | Out-of-zone claims | Out-of-zone $ | Claims | Paid out of zone |
|---|---|---|---|---|
| SD | 54.8% | 54% | 3,690 | $32,474,916 |
| TX | 50.7% | 47.3% | 380,408 | $8,143,547,217 |
| CO | 50.4% | 34.8% | 5,335 | $30,902,497 |
| ND | 49.7% | 60.2% | 11,045 | $141,960,896 |
| MT | 46.1% | 36.7% | 1,598 | $6,211,942 |
| AZ | 40.3% | 42.1% | 4,369 | $22,853,548 |
| ID | 39.7% | 35.3% | 1,031 | $3,564,004 |
| MN | 38.2% | 33.6% | 11,219 | $50,844,817 |
| CA | 35.8% | 29.7% | 46,742 | $214,713,973 |
| NM | 35.7% | 41.7% | 1,718 | $16,981,341 |
| GA | 34.9% | 28.3% | 23,760 | $144,848,863 |
| KS | 33.2% | 31.2% | 6,975 | $34,382,891 |
| NV | 32.9% | 25.4% | 1,788 | $11,765,859 |
| ME | 32.7% | 26.1% | 4,618 | $21,102,683 |
| TN | 32.1% | 26.5% | 16,477 | $125,216,770 |
| OK | 32% | 33.3% | 12,320 | $85,025,841 |
| MD | 30% | 18.9% | 19,931 | $60,414,143 |
| NY | 29.9% | 26.1% | 152,442 | $1,472,137,229 |
| MI | 29.7% | 25.5% | 13,662 | $35,933,763 |
| WI | 29.5% | 29.1% | 8,963 | $39,347,689 |
| NH | 29% | 25% | 3,842 | $15,886,853 |
| PA | 28.8% | 24.9% | 70,464 | $350,601,296 |
| AR | 28.7% | 29.9% | 9,577 | $63,497,509 |
| HI | 28.7% | 27.4% | 5,787 | $42,324,604 |
| OR | 28.7% | 27.3% | 5,761 | $28,582,140 |
| LA | 27.3% | 21.2% | 470,094 | $4,372,137,743 |
| OH | 27.3% | 25.7% | 24,012 | $90,154,003 |
| RI | 26.3% | 17.6% | 6,935 | $24,276,344 |
| MS | 25.6% | 29.5% | 59,311 | $895,006,939 |
| WA | 23.8% | 20.7% | 14,951 | $72,563,220 |
| MA | 23.7% | 21.8% | 31,304 | $90,741,208 |
| IL | 23.6% | 21.2% | 46,527 | $117,607,956 |
| CT | 23.3% | 16.5% | 27,986 | $89,132,231 |
| KY | 22.6% | 19.7% | 24,829 | $98,681,409 |
| VT | 22.5% | 17.1% | 3,515 | $20,052,574 |
| IN | 21.8% | 19.8% | 16,599 | $56,234,359 |
| IA | 21.6% | 25.8% | 14,126 | $88,456,524 |
| NE | 21.5% | 19.9% | 5,574 | $18,924,384 |
| VA | 21.4% | 15.9% | 48,864 | $117,807,972 |
| AL | 20.6% | 25.1% | 43,580 | $295,035,361 |
| WV | 19.8% | 19.4% | 23,080 | $68,630,862 |
| NC | 18.7% | 19% | 107,868 | $427,131,093 |
| SC | 18.1% | 18.6% | 49,144 | $191,150,294 |
| DE | 15.7% | 13.5% | 6,293 | $12,034,377 |
| FL | 14.8% | 8.1% | 444,150 | $1,553,347,022 |
| MO | 14.2% | 13.9% | 48,364 | $131,006,297 |
| NJ | 10.2% | 8.4% | 198,666 | $521,728,935 |
The figures on this page come from two public FEMA datasets and can be rebuilt from them. If you are writing about a specific county, the lookup above gives you its own number.
Questions, corrections, or a request for a custom cut: hello@beforeregret.com. If you find an error in this analysis we would genuinely rather hear it than not.
The full county table, 1,921 rows, one per US county with a classifiable claims history. The same numbers as the lookup above.
outside-the-zone-by-county.csv · the full figures as JSON
Three things to know before you quote the file. The county rows total 2,487,348 classifiable claims rather than the 2,578,413 counted nationally, because not every claim on record carries a county that can be matched; the national and state figures on this page are not built up from the county table. The take-up column is blank for 263 counties rather than zero — blank means no rate is published for that county, including the 22 where the computed rate exceeded 100%, which the source attributes to structure-count errors, and reading a blank as a zero would invert what it means.
And six names appear twice, because they are genuinely two places: Baltimore MD, St. Louis MO, and Fairfax, Richmond, Franklin and Roanoke in Virginia each exist as both an independent city and a separate county. The two rows are different jurisdictions with different claims histories. The file does not say which is which, so if you are writing about one of those six, check the count against the source before you publish rather than picking a row.
Every exhibit on this page as a high-contrast image, free to reuse with credit. The PNG drops straight into a document or a slide; the SVG stays sharp at any size.
Exhibit 1 (PNG) · SVG
Exhibit 2 (PNG) · SVG
More than one paid NFIP flood claim in four came from outside the mapped high-risk flood zone, and in Texas it was about half, according to an analysis of FEMA claims data by BeforeRegret.
Outside the Zone, Before Regret, 1 September 2026. Data source: FEMA NFIP claims · 1,921 counties. https://www.beforeregret.com/research/outside-the-zone/
Free to use on any site. It sizes itself to its content and carries its own credit line.
