Evidence quality 4.88/5
Eight-dimension review score against the quality rubric . Each dimension scored 1–5.
- D1 Source grounding
- 5/5
- D2 Source authority
- 5/5
- D3 Arithmetic
- 5/5
- D4 Uncertainty
- 4/5
- D5 Scope
- 5/5
- D6 Prose
- 5/5
- D7 Perception honesty
- 5/5
- D8 Caveat completeness
- 5/5
● your factors — click this risk ▾ to reveal
- Your factors
≈ As likely as
Perceived
Floods are the deadliest thunderstorm-related hazard in the US and the most frequent natural disaster globally, but we have not found a rigorous recent survey that isolates "fear of being killed in a flood" from general severe-weather or climate anxiety. We mark the perceived side as editorial intuition. Anecdotally, the gap that matters here is not geographic — it is behavioral. The single most consistent finding in the US flood mortality literature is that people routinely underestimate the force of moving water and drive into it; "turn around, don’t drown" exists precisely because the common-sense prior ("it’s only a few inches") is wrong.
Rough estimate: 31.7% of US adults report being afraid or very afraid of a devastating flood (Chapman Survey 2024)
Actual
~6,600 global flood deaths per year (WHO Bulletin, 1990-2022 EM-DAT window)
global
Show derivation
Uses ~6,600 global flood deaths per year as the smoothed long-window average, anchored on the WHO Bulletin analysis of EM-DAT records for 1990-2022 (218,353 total deaths across 4,713 recorded flood events in 168 countries, ≈ 6,617/year). Jonkman’s 1975-2002 analysis gives a comparable ~6,500/year from a different compilation of the same underlying disaster database. Annual per-capita risk ≈ 6,600 / 8,000,000,000 ≈ 8.25e-7; compounded over 60 adult life-years ≈ 4.95e-5, which we display as ~1 in 20,000 global adult lifetime. The window matters: the long-run average is dominated by rare megaevents (the 1931 China floods alone are estimated at 1-4 million deaths, and modern events like the 1998 Bangladesh floods, 2010 Pakistan floods, and 2022 Pakistan floods each killed thousands), while the post-2000 smoothed average is closer to 5,000-7,000 per year. The uncertainty band below brackets the window-sensitivity rather than sampling noise, and the headline is an average-global-adult figure that is essentially meaningless for any individual — see the regional breakdown and caveats.
Caveats: The global-average figure is a scale marker, not a personal estimate. Flood mort…
The global-average figure is a scale marker, not a personal estimate. Flood mortality is heavily concentrated in South Asian and East Asian river deltas, coastal Southeast Asia, and flash-flood-prone terrain in low- and middle-income countries — per-capita lifetime risk in those settings is one to two orders of magnitude above the global average. In high-income countries, where exposure is lower and warning and evacuation infrastructure is mature, the residual risk is dominated by a single behavioral pathway: driving into flooded roadways. The US 30-year record, in which more than half of flood drownings involve vehicles, is the clearest example of a fatality profile that is almost entirely preventable at the margin. "Turn around, don’t drown" is one of the most evidence-supported safety messages on the site because the specific behavior it targets accounts for the majority of the specific deaths it is trying to prevent. The other recurring US fatal mechanism is the nocturnal flash flood: in the early hours of 4 July 2025 a flash flood on the Guadalupe River in Kerr County and surrounding Central Texas killed more than a hundred people — by toll the deadliest US rainfall-driven flash flood in roughly half a century, and one of only a handful since the late 20th century to kill 100 or more. Many victims were asleep when the river rose, including campers at a low-lying summer camp on the floodway, which is the pattern flash-flood mortality keeps repeating: short warning lead time, high water velocity, and exposure at night when evacuation is slowest.
How the risk varies
The headline figure averages across very different situations. Here’s how the probability varies by scenario or context:
1 in 20,202
WHO Bulletin 1990-2022 smoothed annual deaths ÷ 8B × 60-year adult life.
1 in 1,000
Low-lying deltas, monsoon river systems, and extreme single-event tolls (1998 Bangladesh, 2010 and 2022 Pakistan each killed thousands); the 1931 China floods alone are estimated at 1-4 million deaths, illustrating the megaevent dominance of the long-run record.
1 in 50,000
~88 flood deaths per year across a 333M-person country, of which more than half involve vehicles driven into flooded roadways — the behavioral story dominates the US headline.
1 in 10,000
Lower absolute numbers than South Asia but a high mortality fraction per event, driven by limited warning infrastructure and high building vulnerability.
Bar length and shade rank these scenarios against each other, not against other risks. The exact odds are shown beside each.
Related risks
Other risks on similar themes — for exploring related fears.
Drought famine death
What are the odds of dying from drought-induced famine or water scarcity?
Post-hurricane heat death
How likely is dying from heat after a hurricane knocks out your power in summer?
Pick challenger
Across the 33-year window 1990-2022, the WHO Bulletin’s analysis of EM-DAT records attributes 218,353 deaths to floods globally — an average of roughly 6,600 per year, spread across 4,713 recorded events in 168 countries. Jonkman’s earlier 1975-2002 analysis of the same underlying disaster database lands within 2% of that number (~6,500/year from 175,000 deaths over 27 years), which is the main reason we use it as the headline rather than any shorter-window mean. Divided by eight billion people and compounded over a 60-year adult life, the order-of-magnitude figure is about 1 in 20,000 global lifetime — roughly five times higher than the equivalent tsunami figure, about half the global hurricane number, and about an order of magnitude lower than the global earthquake number. Geography dominates: South Asia, East Asia, and coastal Southeast Asia carry most of the absolute mortality, while the WHO European Region accounts for only about 2.5% of the global total.
What makes the US flood story unusual is how much of it is behavioral rather than meteorological. NWS records put the long-run US average near 88 flood deaths per year across a population of 333 million — a lifetime baseline near 1 in 63,000, roughly a third of the global figure. But within that already-low number, more than half of all flood-related drownings involve a vehicle driven into hazardous flood water, according to CDC data cited by NWS. Six inches of fast-moving water can knock an adult off their feet. Twelve inches can float most cars. Twenty-four inches can sweep away SUVs and trucks. The force of moving water is reliably underestimated in real-time, and “turn around, don’t drown” exists because the single most common fatal decision is the one that feels safest: edging into a flooded road to see how deep it is. It is one of the sharpest behavioral multipliers on the site — the lifetime risk for a driver who will, some day, ignore a TADD sign is not the US baseline.
The global trend is the other thing that changes the headline. Flood mortality has declined substantially over the last fifty years despite rising exposure, and Bangladesh is the clearest documented example: cyclone-driven death tolls fell from roughly 500,300 in the 1970 Bhola cyclone to 138,958 in a 1991 cyclone to 4,234 in 2007’s Cyclone Sidr — more than a hundredfold decline in under forty years even as population and floodplain occupation both grew. WHO Bulletin researchers credit the decline to early warning systems, cyclone shelters, evacuation plans, and coastal embankments, not any drop in storm intensity. The long-run average is still pulled hard by megaevents — the 1931 China floods alone are estimated at between one and four million deaths, a figure large enough to move any century-scale denominator — but the post-2000 smoothed annual average is closer to 5,000-7,000 per year than to any of the mid-20th-century peaks. The hazard hasn’t changed. The mortality has.
Related tidbits
Floods kill roughly 6,600 people a year worldwide. In the US the residual risk is dominated by one preventable decision: more than half of flood drownings happen when someone drives into moving water. Six inches can knock an adult off their feet.
Claim ledger
Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.
4/5 sources independently verified verbatim against the cited source
-
[1] Bulletin of the World Health Organization (Liu Q, Du M, Wang Y, Deng J, Yan W, Qin C, Liu M, Liu J) — Global, regional and national trends and impacts of natural floods, 1990-2022 Verified
Global, regional and national trends and impacts of natural floods, 1990-2022- Statistic
4,713 floods recorded in 168 countries between 1990 and 2022 resulted in 218,353 deaths and more than US$1.3 trillion in economic damages; South-East Asia Region had 71,713 deaths (32.84%), Region of the Americas 48,630 (22.27%), Western Pacific Region 42,721 (19.57%), Eastern Mediterranean 29,819, Africa 19,927, Europe 5,543.- Excerpt
“"Between 1990 and 2022, 4713 floods were recorded in 168 countries, which resulted in 218 353 deaths and caused more than US$ 1.3 trillion in economic damages. Of these, the South-East Asia Region had the highest number (71 713; 32.84%) followed by the Region of the Americas (48 630; 22.27%)." ”
- Source data from
- 2024-06-01
- Accessed
- 2026-04-11 · archived copy
- Verification
- Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
- Calculation
- WHO Bulletin total of 218,353 deaths / 33 years ≈ 6,617 deaths per year, which we round to ~6,600 as the headline native figure. Annual per-capita risk ≈ 6,600 / 8,000,000,000 ≈ 8.25e-7; compounded over 60 adult years ≈ 4.95e-5 ≈ 1 in 20,200, which we display as ~1 in 20,000. The regional breakdown in the paper (South-East Asia and Americas together accounting for 55% of deaths, Europe contributing only 2.5%) is the empirical basis for the regional_breakdown rows below.
- Independence
- The WHO Bulletin analysis draws on EM-DAT (CRED) records, the same underlying disaster database used by Jonkman (2005) and by most modern flood mortality research. Treat the two peer-reviewed sources as methodologically independent compilations of overlapping source data — they agree on the order of magnitude (~6,500-6,600 deaths/year) across very different time windows, which is the main reason we use that number rather than a shorter-window mean.
-
[2] Natural Hazards (Jonkman, S.N.) — Global Perspectives on Loss of Human Life Caused by Floods
Global Perspectives on Loss of Human Life Caused by Floods- Statistic
Over the 27-year period studied, more than 175,000 people died and close to 2.2 billion were affected directly by floods worldwide; flash floods produce the highest average mortality per event, and Asian river floods dominate absolute casualty counts.- Excerpt
“"Over 27 years, more than 175,000 people died and close to 2.2 billion were affected directly by floods worldwide. ... Flash floods result in the highest average mortality per event. ... On a worldwide scale Asian river floods are most significant in terms of number of persons killed and affected." ”
- Source data from
- 2005-01-01
- Accessed
- 2026-04-11 · archived copy
- Calculation
- Jonkman’s 175,000 deaths / 27 years ≈ 6,500/year — within 2% of the WHO Bulletin figure despite covering a different window (1975-2002 vs 1990-2022) and using a different compilation of the EM-DAT records. Jonkman is the source traditionally cited for the "flash floods have the highest per-event mortality" and "Asian river floods dominate global casualty counts" claims, both of which shape the regional_breakdown and body text here.
- Independence
- Jonkman (2005) and Liu et al (2024) both draw ultimately from EM-DAT / CRED disaster records. Treated as methodologically independent because of the different time windows, different aggregation choices, and different research groups; their agreement on order of magnitude is the main quantitative anchor for the headline.
-
[3] NOAA National Weather Service — Turn Around Don't Drown — Flood Safety Verified
Turn Around Don't Drown — Flood SafetySee all 2 Likelier entries citing this source →
- Statistic
Per CDC data cited by NWS, over half of all US flood-related drownings occur when a vehicle is driven into hazardous flood water. 6 inches of moving water can knock an adult off their feet; 12 inches of moving water can carry away most cars; 24 inches can sweep away SUVs and trucks.- Excerpt
“"According to the Centers for Disease Control and Prevention, over half of all flood-related drownings occur when a vehicle is driven into hazardous flood water. ... Six inches of fast-moving water can knock over an adult. It takes just 12 inches of rushing water to carry away most cars and just 2 feet of rushing water can carry away SUVs and trucks." ”
- Source data from
- 2024-01-01
- Accessed
- 2026-04-11 · archived copy
- Verification
- Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
- Calculation
- Used to anchor the US behavioral-multiplier story rather than the global normalized figure. The NWS / CDC "over half" framing is the evidence base for the personal_factor_multipliers entry on driving into flooded roadways and for the "turn around, don’t drown" paragraph in the body text. NWS reports an approximate long-run average near 88 US flood deaths per year across its 30-year window, which gives a US lifetime baseline of ~88/333M × 60 ≈ 1.6e-5 ≈ 1 in 63,000 — an order of magnitude lower than the global figure because US flood mortality is dominated by behavioral (vehicle) rather than exposure (coastal-delta inundation) pathways, and because US warning and evacuation infrastructure is unusually mature.
- Independence
- NWS hazard-statistics pages and CDC mortality reporting are partially overlapping (NWS Storm Data is one input CDC uses to classify weather-related deaths), so the two figures should be treated as a single US authoritative chain rather than two fully independent estimates.
-
[4] NOAA National Environmental Satellite, Data, and Information Service (NESDIS) — NOAA Satellites Inform Warning for the Texas Hill Country Floods Verified
NOAA Satellites Inform Warning for the Texas Hill Country Floods- Statistic
In the early hours of 4 July 2025 a devastating flash flood occurred in the Central Texas Hill Country in which more than a hundred people lost their lives and thousands of residents' homes were destroyed — by toll the deadliest US rainfall-driven flash flood in roughly half a century.- Excerpt
“"In the early hours of July 4, 2025 a devastating flash flood occurred in the Central Texas Hill Country. ... more than a hundred people lost their lives and thousands of residents' homes were destroyed." ”
- Source data from
- 2025-07-01
- Accessed
- 2026-06-14 · archived copy
- Verification
- Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
- Calculation
- Not used to change the headline normalized figure — it is a single high-fatality US event that does not shift the long-run global per-capita average. It is cited as a current data point reinforcing the nocturnal flash-flood subgroup discussed in the caveats and the flash-flood personal_factor_multipliers entry: the predawn timing, short warning lead time, and exposure of sleeping victims on a low-lying floodway are the same failure modes Jonkman (2005) identifies as producing the highest per-event mortality fraction. The toll ("more than a hundred"; ~138-139 across Kerr, Travis, Burnet, Williamson and Tom Green counties per contemporaneous reporting) makes it the deadliest US rainfall-driven flash flood since the 1976 Big Thompson Canyon flood.
- Independence
- NOAA/NWS post-event analyses of the July 2025 Texas flood share the same federal weather-service reporting chain as the Turn Around Don't Drown / NWS US mortality figures above; treat as one US authoritative chain, not an independent estimate.
-
[5] Bulletin of the World Health Organization (Haque U, Hashizume M, Kolivras KN, Overgaard HJ, Das B, Yamamoto T) — Reduced death rates from cyclones in Bangladesh: what more needs to be done? Verified
Reduced death rates from cyclones in Bangladesh: what more needs to be done?- Statistic
Cyclone-related death tolls in Bangladesh fell from roughly 500,300 in the 1970 Bhola cyclone to 138,958 in a 1991 cyclone to 4,234 in 2007's Cyclone Sidr, a more-than-hundredfold decline credited to early warning systems, cyclone shelters, evacuation plans, and coastal embankments.- Excerpt
“"The main factors responsible for these reduced fatalities and injuries are improved defensive measures, including early warning systems, cyclone shelters, evacuation plans, coastal embankments, reforestation schemes and increased awareness and communication." ”
- Source data from
- 2012-02-01
- Accessed
- 2026-07-04 · archived copy
- Verification
- Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
- Calculation
- Supplies the evidentiary basis for the Bangladesh mortality-decline paragraph in the body text. Death tolls of roughly 500,300 (1970 Bhola cyclone), 138,958 (1991 cyclone), and 4,234 (2007 Cyclone Sidr) represent more than a hundredfold reduction over less than four decades, which the paper attributes to early warning systems, cyclone shelters, evacuation plans, and coastal embankments, not any reduction in storm intensity. Cyclone-driven storm-surge flooding is the dominant flood-fatality pathway in coastal Bangladesh, so this is used as the evidence for the "mortality has declined" thesis in the body text, distinct from the river-flood events (1998, 2010, 2022) discussed elsewhere in this entry.
- Independence
- Published in the same journal (Bulletin of the WHO) as the Liu et al. 2024 global flood analysis above but a different research group studying a different, narrower population (Bangladesh cyclone mortality specifically, not global flood mortality); treat as independent of the other sources here.