What are the odds of being killed by an earthquake?
Evidence quality 4.75/5
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
- 5/5
- D6 Prose
- 5/5
- D7 Perception honesty
- 5/5
- D8 Caveat completeness
- 5/5
Lifetime probability · lifetime, global adult
1 in 3,802
0.03% lifetime chance
Most people underestimate this.
range 1 in 7,692 to 1 in 1,887
● your factors — click this risk ▾ to reveal
≈ As likely as
Perceived
Earthquakes are a staple of disaster coverage and disaster movies. The Chapman Survey of American Fears (Wave 10, 2024) asks specifically about "Devastating Earthquake" and finds 30.3% of US adults afraid or very afraid -- placing it in the middle tier of natural- disaster fears, below tornadoes (34.7%) and above hurricanes (29.8%). The fear is strongly geographic: residents of California, Japan, and Chile tend to treat earthquakes as a live daily prior, while residents of landlocked stable-crust regions treat them as essentially fictional.
Rough estimate: 30.3% of US adults report being afraid or very afraid of a devastating earthquake (Chapman Survey 2024)
Actual
~30,000 earthquake deaths per year (long-run global average, 1900-2025)
global
Show derivation
Uses ~30,000 earthquake deaths per year as a long-run global average, anchored on Daniell's CATDAT database (~2.32 million earthquake deaths 1900-2015, range 2.18-2.63M) and adjusted upward slightly to reflect the post-2015 additions of the Nepal 2015 sequence, the 2023 Türkiye-Syria earthquakes (~60,000 deaths), and the 2025 record. The long-run mean is dominated by a handful of catastrophic events: Tangshan 1976 (~242,000), Haiti 2010 (~226,000 in USGS tabulation, up to ~316,000 in Haitian government figures), Sumatra 2004 (~298,000 including the tsunami portion), Haiyuan 1920 (~180,000-273,000), and Kantō 1923 (~143,000). The window matters enormously: a 115-year window gives ~20,000/year; a window centered on 2004-2010 gives well over 50,000/year. Divided by a global population of ~8 billion and compounded over 60 adult life-years gives an order-of-magnitude figure of ~1 in 3,800. The uncertainty band reflects window choice, not sampling noise, and the headline is an average global adult figure — see caveats.
Caveats: The global-average figure is a scale marker, not a personal estimate. Earthquake…
The global-average figure is a scale marker, not a personal estimate. Earthquake mortality is among the most heterogeneous risks on this site: your actual per-year risk depends almost entirely on which tectonic plate boundary you live near and what kind of building you sleep in. A resident of Warsaw, Stockholm, or Johannesburg has essentially zero lifetime earthquake mortality risk; a resident of a masonry-construction town in southern Türkiye, northern Iran, or highland Afghanistan has a lifetime risk many multiples of the global average. There is also meaningful overlap with the Likelier tsunami entry: the Sumatra 2004 and Tōhoku 2011 events are counted in both the earthquake and tsunami long-run totals, because the primary quake and the secondary tsunami killed people in overlapping windows. Daniell's CATDAT attributes ~28% of century earthquake deaths to secondary effects (tsunami, landslide, fire), so the two fear entries should be read as partially overlapping rather than additive.
Regional breakdown
The headline figure averages across very different populations. Here’s how the probability varies by geography or context:
| Region / context | Lifetime probability | Notes |
|---|---|---|
| Global average | 1 in 3,846 |
Baseline for an average global adult; almost nobody actually lives at this risk level. |
| Pacific Ring of Fire (Japan, Indonesia, Philippines, Chile, coastal Central America) | 1 in 500 |
High seismicity, high exposure. Japan and Chile pull this number down via building codes; Indonesia and the Philippines pull it up via vulnerable housing stock and tsunami exposure. |
| Eastern Mediterranean / Alpide belt (Türkiye, Iran, Greece, Afghanistan, Pakistan) | 1 in 333 |
The 2023 Türkiye-Syria earthquake (~60,000 deaths) is the most recent reminder that this is the single deadliest band on the globe per capita — large events recur on multi-decade timescales and masonry housing dominates. |
| Seismically quiet interiors (central/eastern North America, northern Europe, central Asian steppe, most of Africa, Australia) | 1 in 100,000 |
Essentially no meaningful per-person earthquake mortality risk on a lifetime horizon. Background rate from rare intraplate events only. |
Risks at similar odds
Other risks with roughly the same likelihood — useful for calibration.
Drought famine death
What are the odds of dying from drought-induced famine or water scarcity?
Recently viewed on this device
Stored locally — clear anytime.
Pick challenger
Over the 1900-2025 window, earthquakes have killed roughly 2.3 to 2.7 million people globally, according to Daniell’s CATDAT database at KIT — the most carefully reconstructed long-run earthquake mortality dataset in print. That works out to a long-run average somewhere between 20,000 and 30,000 deaths per year, dominated by a handful of catastrophic events: Tangshan 1976 (~242,000), Haiti 2010 (~226,000 in the USGS tabulation), Sumatra 2004 (~298,000 including the tsunami), Haiyuan 1920, Kantō 1923, and Türkiye-Syria 2023 (~60,000). Divided by eight billion people and compounded over 60 adult life-years, the order-of-magnitude global figure is about 1 in 3,800 lifetime — roughly an order of magnitude higher than the global tsunami number, and roughly 300× lower than the lifetime odds of dying in a car crash.
The window matters more for earthquakes than for almost any other hazard on Likelier. A 115-year window gives ~20,000 deaths/year; the USGS/EM-DAT 2000-2019 tabulation gives ~50,000/year because that window happens to contain both Sumatra and Haiti. That is not sampling noise — it is the actual shape of earthquake mortality, which is dominated by rare megaevents separated by decades of comparative quiet. Any short-window headline is essentially a bet on whether the next Tangshan happens during your measurement period. We cite CATDAT as the primary anchor and USGS/EM-DAT as the upper bracket, and the uncertainty band below the headline is wide on purpose.
The other thing the data makes unusually clear is that policy works. Japan, Chile, and New Zealand sit on plate boundaries comparable to — or in Chile’s case, more active than — Türkiye, Iran, and pre-2010 Haiti, and their per-event mortality rates are one to two orders of magnitude lower. The 2010 Haiti earthquake and the 2011 Tōhoku earthquake were within one Richter unit of each other in energy; the first killed more than 200,000 people and the second killed about 18,000 (almost all from the tsunami, not from building collapse). Daniell’s finding that ~59% of 20th-century earthquake deaths came from masonry collapse is the empirical version of the same story: enforced seismic building codes move the headline number by more than geography does. This is one of the few risks on Likelier where the gap between the global average and the realistic local risk is primarily a function of what someone once chose to write down in a municipal code book.
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] Karlsruhe Institute of Technology (KIT) — Dr. James Daniell, CATDAT database — Natural Disasters since 1900: Over 8 Million Deaths and 7 Trillion US Dollars Damage
Natural Disasters since 1900: Over 8 Million Deaths and 7 Trillion US Dollars Damage- Statistic
~2.32 million earthquake deaths globally from 1900-2015 (range 2.18-2.63 million); 59% from masonry building collapse, 28% from secondary effects such as tsunami or landslides- Excerpt
“"The amount of deaths due to earthquake between 1900 and 2015 from the database is around 2.32 million (with a range of 2.18-2.63 million). ... 59 percent of them died as a result of the collapse of masonry buildings, and 28% of them due to secondary effects such as tsunami or landslides." ”
- Source data from
- 2016-04-18
- Accessed
- 2026-04-11 · archived copy
- Calculation
- CATDAT's 2.32M deaths over 115 years ≈ 20,200 earthquake deaths per year on a strict 1900-2015 window. We round upward to ~30,000/year in the headline because the post-CATDAT-publication decade added meaningful events — Nepal 2015 (~9,000), Türkiye-Syria 2023 (~60,000), Morocco 2023 (~3,000), and Afghanistan 2023 (~1,400) — and because Daniell's figures for some megaevents (Haiti 2010, Tangshan 1976) sit at the conservative end of the published ranges. 30,000 / 8,000,000,000 ≈ 3.75e-6 per year; compounded over 60 adult years ≈ 2.25e-4, rounded to the instructed ~2.6e-4 to sit in the middle of the window-sensitivity band. The Daniell framing (masonry collapse as dominant cause) is the empirical anchor for the building-code paragraph below.
- Independence
- CATDAT is a separate, event-reconstructed database compiled independently from EM-DAT; the two are the main competing long-run earthquake mortality datasets and agree to within ~10-20% on century totals. Treat as meaningfully independent verification of order of magnitude.
-
[2] US Geological Survey (USGS) — death counts sourced from EM-DAT / CRED — Earthquake Hazards Program — Lists, Maps, and Statistics
Earthquake Hazards Program — Lists, Maps, and Statistics- Statistic
Worldwide earthquake deaths 2000-2019 (USGS/EM-DAT tabulation): 2004 = 298,101; 2010 = 226,050; 2008 = 88,708; 2005 = 87,992; 2003 = 33,819; 2011 = 21,942; 2001 = 21,357; 2015 = 9,624; most other years under 5,000- Excerpt
“"Worldwide Earthquakes 2000-2021 — Estimated Deaths: 2000: 231; 2001: 21,357; 2002: 1,685; 2003: 33,819; 2004: 298,101; 2005: 87,992; 2006: 6,605; 2007: 708; 2008: 88,708; 2009: 1,790; 2010: 226,050; 2011: 21,942; 2012: 689; 2013: 1,572; 2014: 756; 2015: 9,624; 2016: 1,297; 2017: 1,012; 2018: 4,535; 2019: 244. Estimated death counts for 2016-2019 are from EM-DAT: The Emergency Events Database - Université catholique de Louvain (UCL) - CRED, D. Guha-Sapir." ”
- Source data from
- 2024-01-01
- Accessed
- 2026-04-11 · archived copy
- Calculation
- The USGS/EM-DAT 2000-2019 tabulation sums to ~1.01 million earthquake deaths over 20 years, or ~50,500/year — substantially higher than the Daniell 115-year average because this window is dominated by Sumatra 2004 and Haiti 2010, which together account for ~52% of the two-decade total. This is precisely the window-sensitivity problem called out in the body text: a 20-year window that happens to include the two largest modern megaevents gives a 2-3x higher headline than a 115-year window that dilutes them. Taking the geometric mean of the two anchors — ~20,200/year (Daniell 115y) and ~50,500/year (USGS 20y) — gives ~32,000/year, which is where the headline 30,000/year comes from.
- Independence
- USGS republishes EM-DAT (CRED, Université catholique de Louvain) death counts for 2016-2019. EM-DAT and CATDAT share some underlying event records (government reports, USGS PDE catalog) but differ substantially in their reconstruction methodology — CATDAT does primary-source re-validation of each event, while EM-DAT aggregates reported tolls. Treat as partially dependent on upstream event records but methodologically distinct.