What are the odds of being killed by a tornado?
Evidence quality 4.5/5
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
- 4/5
- D4 Uncertainty
- 4/5
- D5 Scope
- 4/5
- D6 Prose
- 4/5
- D7 Perception honesty
- 5/5
- D8 Caveat completeness
- 5/5
Lifetime probability · lifetime, US adult
1 in 80,645
0.001% lifetime chance
Most people overestimate this.
range 1 in 142,857 to 1 in 50,000
● your factors — click this risk ▾ to reveal
≈ As likely as
Perceived
Fear of tornadoes is strongly geographic — survey work on tornado anxiety is concentrated in the Plains and the Southeast, and there is no widely cited national poll that isolates "fear of being killed by a tornado" from general severe-weather anxiety. We mark the perceived side as editorial intuition. Anecdotally, residents of Dixie Alley and Tornado Alley tend to overestimate per-year risk in low-activity years and underestimate it in outbreak years; residents of the Northeast and West Coast tend to treat tornadoes as a near-zero hazard, which for them is roughly accurate.
Rough estimate: 34.7% of US adults report being afraid or very afraid of a devastating tornado (Chapman Survey 2024)
Actual
~70 tornado fatalities per year in the US (long-run average)
US residents
Show derivation
Uses ~70 US tornado fatalities per year as a long-run average that smooths out the extreme year-to-year variability documented by NOAA SPC and the Insurance Information Institute (annual totals between 2014 and 2024 ranged from 10 in 2018 to 103 in 2021). Divides by US population (~333M) and compounds over 59 years of remaining adult life. The lifetime figure is a national average and is essentially meaningless for any specific resident — see caveats.
Caveats: The national-average lifetime figure (~1 in 80,000) is almost meaningless for an…
The national-average lifetime figure (~1 in 80,000) is almost meaningless for any individual reader. Tornado fatalities cluster heavily in the Southeast (Dixie Alley: Alabama, Mississippi, Tennessee, Arkansas) and the central Plains (Tornado Alley: Oklahoma, Kansas, Missouri, Texas). Ashley's climatology work shows that Dixie Alley has historically produced more tornado fatalities than Tornado Alley despite fewer total tornadoes, largely because Dixie Alley storms are more often nocturnal, rain-wrapped, and strike denser, more vulnerable housing stock. A resident of central Mississippi faces a per-year risk many times the national average; a resident of Vermont, Oregon, or coastal California faces a risk close to zero. Risk also depends heavily on housing type — manufactured-home residents account for a disproportionate share of fatalities relative to their share of the population.
Risks at similar odds
Other risks with roughly the same likelihood — useful for calibration.
AMOC collapse
What are the odds of the AMOC experiencing an abrupt collapse before the end of your lifetime?
Recently viewed on this device
Stored locally — clear anytime.
Pick challenger
Across the US, an average of roughly 70 people per year die in tornadoes, drawing on the long-run NOAA Storm Prediction Center record. Divided across the population and compounded over an adult lifetime, that comes to roughly 1 in 80,000 — a number remarkably close to the lifetime odds of being killed by lightning, and about three orders of magnitude lower than the lifetime odds of dying in a car crash.
What makes tornado fatalities unusual among hazards is how violently the annual count moves around. The 2014-2024 window contains a year with 10 deaths (2018) and a year with 103 (2021), and the longer record contains the 2011 super-outbreak, which alone killed 553 people. A single bad April afternoon in Alabama or Missouri can move the decade average. That makes any short-window estimate suspect, and it makes “deaths per year” a much noisier statistic than, say, drowning or lightning, where the year-to-year totals are nearly stationary.
The bigger problem with the national figure is that it averages over a country where the underlying risk varies by more than an order of magnitude across geography. Ashley’s climatology work on Dixie Alley shows that the Southeast — Alabama, Mississippi, Tennessee, Arkansas — has historically produced more tornado fatalities than the classic Plains “Tornado Alley” despite a lower total tornado count, because Dixie Alley storms are more often nocturnal, rain-wrapped, and aimed at more vulnerable housing. A resident of central Mississippi and a resident of coastal Maine are both included in the 1-in-80,000 number, and it is roughly accurate for neither of them.
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.
-
[1] NOAA / NWS Storm Prediction Center — SPC Warning Coordination Meteorologist Page — Tornado, Hail, and Wind Statistics
SPC Warning Coordination Meteorologist Page — Tornado, Hail, and Wind Statistics- Statistic
30-year (1996-2025), 20-year (2006-2025), and 10-year (2016-2025) state-level annual averages of US tornado fatalities, derived from the SPC tornado database (1950-2024)- Excerpt
“"Annual Averages: Tornado Fatalities by State — 30 Year Average (1996-2025), 20 Year Average (2006-2025), 10 Year Average (2016-2025). The SPC tornado database CSV files cover 1950 through 2024." ”
- Source data from
- 2026-01-12
- Accessed
- 2026-04-11 · archived copy
- Calculation
- The SPC publishes the underlying yearly fatality counts and state-level averages we use to anchor the long-run ~70/year national figure. SPC's per-state maps are also what makes the geographic non-uniformity quantitative — fatalities are heavily concentrated in Alabama, Mississippi, Tennessee, Arkansas, Missouri, and the Plains states.
- Independence
- SPC is the upstream source for most other US tornado fatality datasets, including the Insurance Information Institute table cited below, so these two sources should be treated as a single authoritative chain rather than two independent estimates.
-
[2] Insurance Information Institute (data sourced from NOAA SPC / NWS) — Facts + Statistics: Tornadoes and Thunderstorms
Facts + Statistics: Tornadoes and Thunderstorms- Statistic
US tornado fatalities by year, 2014-2024: 47, 36, 18, 35, 10, 42, 76, 103, 23, 83, 54 (mean ≈ 48/year for that window)- Excerpt
“"Tornadoes and Related Deaths in the United States, 2006-2025. 2014: 47; 2015: 36; 2016: 18; 2017: 35; 2018: 10; 2019: 42; 2020: 76; 2021: 103; 2022: 23; 2023: 83; 2024: 54. Source: U.S. Department of Commerce, Storm Prediction Center, National Weather Service." ”
- Source data from
- 2025-12-01
- Accessed
- 2026-04-11 · archived copy
- Calculation
- The 2014-2024 window averages ~48/year, but is depressed by several unusually quiet years (2016, 2018, 2022). Longer windows that include the 2011 super-outbreak year (553 deaths) push the long-run average closer to 70/year, which is why we use 70 as the headline native figure rather than the most recent decade's mean. Either choice gives a lifetime risk between 1 in 80,000 and 1 in 120,000.
- Independence
- I.I.I. republishes NOAA SPC / NWS data verbatim. Treated as a convenient year-by-year tabulation, not as independent verification.
-
[3] NOAA National Severe Storms Laboratory (Harold Brooks) — A Brief History of Deaths from Tornadoes in the United States
A Brief History of Deaths from Tornadoes in the United States- Statistic
Mean annual US tornado death toll fell from ~260 (1912-1936) to ~54 (1976-2000); per-capita rate dropped roughly 15-fold from ~1.7 per million in the early 20th century to ~0.12 per million by 2000- Excerpt
“"The mean annual death toll [in 1912-1936] was 260, almost five times as many as in 1976-2000, when the mean was 54. ... Death rates remained relatively constant at approximately 1.6-1.8 per million population annually [pre-1925]. After 1925, rates declined sharply to roughly 0.12 per million by 2000 — approximately a 15-fold decrease." ”
- Source data from
- 2009-03-01
- Accessed
- 2026-04-11 · archived copy
- Calculation
- Brooks's long-run NSSL analysis is the basis for the claim that the modern era is qualitatively different from the early-20th-century baseline. We use his per-capita figure (~0.12 per million ≈ 0.012 per 100k) as a sanity check on our 70/333M ≈ 0.021 per 100k headline; the small drift upward reflects population growth and the inclusion of high-outbreak years like 2011.
- Independence
- Brooks's analysis predates SPC's most recent decade of data and uses NWS Storm Data directly, so it provides genuine methodological independence on the historical trend even though the underlying database is shared.