βš™οΈ How-To Tsunamis Natural hazards & risk Notebook By

How to Rank the Deadliest Tsunamis

On this page

Problem statement

The Indian Ocean tsunami of 26 December 2004 killed more people than any other in the record, and a few others, from Lisbon in 1755 to Japan in 2011, are famous for their tolls. Ranking them from NOAA's database takes three decisions. Use the tsunami's own deaths, not the event total, which includes deaths from the earthquake or eruption. Keep only events that certainly happened. Fill in the old events that record only a rough size of the toll, a code from 1 to 4. This page ranks the deadliest tsunamis and counts how many people tsunamis have killed by century.

Quick answer

Rank by deaths among events of validity 3 or 4:

sure = ev[ev.eventValidity >= 3]
top = sure.nlargest(10, "deaths")[["year", "locationName", "country", "causeCode", "eqMagnitude", "deaths"]]
print(top.to_string(index=False))
 year                 locationName     country  causeCode  eqMagnitude   deaths
 2004      OFF W. COAST OF SUMATRA   INDONESIA        1.0          9.1 227899.0
 1755                       LISBON    PORTUGAL        1.0          8.5  50000.0
 1883                     KRAKATAU   INDONESIA        6.0          NaN  34417.0
 1896                      SANRIKU       JAPAN        1.0          8.3  27122.0
 1868               NORTHERN CHILE       CHILE        1.0          8.5  25000.0
 2011                HONSHU ISLAND       JAPAN        1.0          9.1  18423.0
 1792 SHIMABARA BAY, KYUSHU ISLAND       JAPAN        5.0          6.4  15000.0
 1771               RYUKYU ISLANDS       JAPAN        1.0          7.4  13486.0
 1952                    KAMCHATKA      RUSSIA        1.0          9.0  10000.0
 1976                    MORO GULF PHILIPPINES        1.0          8.0   6800.0

The Indian Ocean tsunami of 2004 killed 227,899 people, more than four times the next. Lisbon's 1755 tsunami killed 50,000, and the Krakatau eruption of 1883 34,417. Japan has four of the ten: Sanriku 1896, Tōhoku 2011, the Ryukyu Islands 1771, and Shimabara 1792, when an earthquake brought down a mountain beside Unzen volcano into the sea. All but two were raised by earthquakes of magnitude 7.4 or more.

Horizontal bars on a log scale of the ten deadliest tsunamis, from Sumatra 2004 with 227,899 to the Moro Gulf 1976 with 6,800. Horizontal bars on a log scale of the ten deadliest tsunamis, from Sumatra 2004 with 227,899 to the Moro Gulf 1976 with 6,800.
The ten deadliest tsunamis.

What goes into a death toll

1. Deaths are known for few events

print(f"events with a death count: {sure.deaths.notna().sum()} of {len(sure):,}; with only a size code: {(sure.deaths.isna() & sure.deathsAmountOrder.notna()).sum()}")
events with a death count: 244 of 1,699; with only a size code: 73

Only 244 of the 1,699 probable or definite tsunamis have a death count, and 73 more have only a size code, from "a few" to "over 1,000". The rest record no deaths, either because there were none or because nobody counted. Any total is a minimum, especially before 1900.

2. The 2004 toll by country

s04 = ev[(ev.year == 2004) & (ev.month == 12) & (ev.day == 26)].iloc[0]
print(ru[ru.tsunamiEventId == s04.id].groupby("country").deaths.sum().nlargest(6).to_string())
country
INDONESIA          167540.0
SRI LANKA           35322.0
INDIA               26047.0
THAILAND             8212.0
SOMALIA               572.0
MYANMAR (BURMA)       150.0

Assigned to the places where they happened, the 2004 deaths fall in Indonesia (167,540), Sri Lanka (35,322), India (26,047) and Thailand (8,212), and as far away as Somalia, across the Indian Ocean. The event itself is filed under Indonesia, where the earthquake was, a trap the country page deals with.

Horizontal bars on a log scale of 2004 tsunami deaths by country: Indonesia 167,540, Sri Lanka 35,322, India 26,047, Thailand 8,212 and Somalia 572. Horizontal bars on a log scale of 2004 tsunami deaths by country: Indonesia 167,540, Sri Lanka 35,322, India 26,047, Thailand 8,212 and Somalia 572.
Where the 2004 tsunami killed people.

3. Recorded deaths by century

c = sure[sure.year >= 1500]
print(c.groupby(c.year // 100 * 100).deaths.agg(["count", "sum"]).astype(int).to_string())
      count     sum
year               
1500      6    4066
1600     14   21889
1700     26  104005
1800     33   99851
1900    114   44183
2000     42  252964

By century, the 18th has 104,005 recorded tsunami deaths, the 19th 99,851 and the 20th 44,183. The 21st already has 252,964 from 42 deadly events, nearly all from 2004 and 2011. Tsunami deaths come in rare catastrophes, so a total over a century is dominated by its largest event.

Deadliest by cause and region

The data used on this page

NOAA's Global Historical Tsunami Database from the National Centers for Environmental Information (doi:10.7289/V5PN93H7), public domain: every event and run-up from its HazEL API, downloaded on 19 October 2026 and saved on this site, because records are added and revised. Heights are in metres and distances in kilometres.

from pathlib import Path
import json
import zipfile

import geopandas as gpd
import numpy as np
import pandas as pd
import requests

Path("data").mkdir(exist_ok=True)
# NOAA NCEI Global Historical Tsunami Database (doi:10.7289/V5PN93H7): every event and every run-up from the HazEL API,
# with Natural Earth's 1:50m land and 1:110m countries. Saved 19 October 2026: events and run-ups are added and revised.
# Live: https://www.ngdc.noaa.gov/hazel/hazard-service/api/v1/tsunamis/events?page=1 (and .../runups), 200 a page
ARCHIVE = "ncei-tsunami-2026-10-19.zip"
if not Path(f"data/{ARCHIVE}").exists():
    Path(f"data/{ARCHIVE}").write_bytes(requests.get(f"https://data.spatialworkflow.io/{ARCHIVE}", timeout=600).content)
Z = zipfile.ZipFile(f"data/{ARCHIVE}")
ev = pd.DataFrame(json.loads(Z.read("tsunami-events.json")))
ru = pd.DataFrame(json.loads(Z.read("tsunami-runups.json")))
print(f"NCEI tsunami database: {len(ev):,} events, {len(ru):,} run-ups")
NCEI tsunami database: 3,138 events, 35,129 run-ups

Example 1: by cause

names = {1: "earthquake", 3: "earthquake and landslide", 5: "volcano, earthquake and landslide", 6: "volcano", 7: "volcano and landslide", 8: "landslide", 9: "meteorological"}
print(sure.groupby("causeCode").deaths.sum().rename(index=names).sort_values(ascending=False).head(5).astype(int).to_string())
causeCode
earthquake                           465337
volcano                               35244
earthquake and landslide              22447
volcano, earthquake and landslide     15000
landslide                              4311

By cause, earthquakes account for 465,337 deaths, volcanoes 35,244, mostly Krakatau, and earthquakes with landslides 22,447. The 15,000 from an earthquake, volcano and landslide together are the single Shimabara disaster. Landslide tsunamis are among the tallest waves, but they are local and have killed 4,311.

Bars of recorded tsunami deaths by century since 1500, about 100,000 each in the 18th and 19th centuries, 44,183 in the 20th and 252,964 so far in the 21st. Bars of recorded tsunami deaths by century since 1500, about 100,000 each in the 18th and 19th centuries, 44,183 in the 20th and 252,964 so far in the 21st.
Recorded tsunami deaths by century.

Example 2: deaths since 1900 by decade

d = sure[sure.year >= 1900]
print(d.groupby(d.year // 10 * 10).deaths.sum().astype(int).to_string())
year
1900      5547
1910       265
1920       758
1930      3350
1940      4932
1950     10062
1960      5776
1970      9549
1980       133
1990      3811
2000    229122
2010     23828
2020        14

Since 1900 the decades vary by orders of magnitude, from 14 deaths so far in the 2020s and 133 in the 1980s to 229,122 in the 2000s. Without the 2004 tsunami the 2000s would have about 1,200. A trend in tsunami deaths needs very long periods, or a measure that isn't dominated by one event, such as the number of deadly tsunamis per decade.