1,699 tsunamis from NOAA's Global Historical Tsunami Database, from the eruption of Thera to 2026: what set them off, the 33,176 places where the water was seen, how high it ran, how far it crossed the ocean, and the people it killed.
A tsunami is the sea moving all at once. Something shoves a huge volume of water, usually the sea floor jerking up in an earthquake, sometimes a volcano or a mountainside falling into a bay, and the push spreads out as waves that can cross an ocean in a day. Near the source they can be tens of metres high. Far away they may be a few centimetres on a tide gauge, or a surge that still drowns a harbour.
NOAA's National Centers for Environmental Information (NCEI) keep a list of every one they can find, from geology and ancient texts to deep-ocean pressure sensors. I took the whole database, 3,138 possible tsunami sources and 35,129 records of where the water arrived, and kept the 1,699 that NCEI rates as definite (1,110) or probable (589). Press play on the map to watch the record build up, or take the tour of the famous waves. Click any source to draw lines to every place its water reached.
Large circles: tsunami sources NCEI rates definite or probable (1,583 of the 1,699 have a position), coloured by cause and sized by the deaths NCEI records for the tsunami (or by the maximum water height). Small dots: the 32,628 places with a position where the water was recorded, coloured by height. Click a source to join it to its places: the lines are straight joins, not the path the wave took. The time scale stretches the recent past, because most of the record is recent. Source: NOAA NCEI/WDS Global Historical Tsunami Database, doi:10.7289/V5PN93H7 (public domain). Outlines: Natural Earth. Plate boundaries: PB2002 (Bird 2003; GeoJSON by H. Ahlenius). Sea floor tiles © Esri, GEBCO, NOAA.
Of the 1,699 tsunamis, 1,224 (72%) were started by an earthquake alone, and 115 (6.8%) by an earthquake that also set off a landslide, so 79% involve an earthquake. These are mostly big ones: the median magnitude among the 1,219 with one recorded is 7.1, and 7 reached magnitude 9 or more. They cluster on the edges of the tectonic plates: 63% of earthquake sources lie within 100 km of a plate boundary in Peter Bird's model, and 92% within 250 km.
What starts a tsunami, and what kills
Share of the 1,699 definite and probable tsunamis by NCEI's cause code, grouped (left), and share of the 545,939 recorded tsunami deaths (right). Hover a bar.
The rest are stranger. Landslides account for 131 (7.7%), many of them rock falls into fjords and lakes. Volcanoes account for 103 (6.1%), from the collapse of Thera in the Bronze Age to Krakatau in 1883 and Hunga Tonga in 2022. 76 are meteotsunamis, waves pushed by fast-moving storms and pressure jumps rather than by the ground. Two were explosions: the Halifax munitions ship in 1917 and the blasting of Ripple Rock, a reef in a Canadian channel, in 1958. Volcanoes cause few tsunamis but deadly ones: 10% of all recorded tsunami deaths, mostly Krakatau and the collapse of Mount Unzen's flank into Shimabara Bay in 1792.
The Pacific is ringed by subduction zones, where one plate dives under another and stores up the strain that great earthquakes release. 1,142 tsunamis (67%) started in NCEI's Pacific regions, and 1,131 (67%) on the Pacific rim itself, the Ring of Fire, leaving out Hawaii. Japan alone accounts for 304 (18%). The Mediterranean, with its long written history, has 232 (14%); the Atlantic and the Caribbean together 197 (12%); the Indian Ocean 98 (5.8%).
Where tsunamis start, and where people died
Share of tsunamis (left) and of recorded tsunami deaths (right) by ocean, from NCEI's region of the source. Hover a bar.
Deaths do not follow the count. The Indian Ocean has only 5.8% of the tsunamis but 50% of the recorded deaths, almost all from a single morning in 2004. The Pacific, with two thirds of the tsunamis, has 38%.
NCEI records a water height for 29,343 places. Most are modest: the median is 2.1 m, and 36% are under a metre. But the record heights are not from the great ocean tsunamis at all. They come from landslides into narrow bays. On 9 July 1958 (10 July in universal time) a magnitude 7.8 earthquake shook a mass of rock into Lituya Bay in Alaska, and the water ran 524.3 m up the opposite slope, stripping the forest. In 2025 a rockslide into Tracy Arm, another Alaskan fjord, sent water 481 m up its walls. All 9 tsunamis with a runup of 100 m or more were set off by a landslide, a volcano, or both.
The highest water on record
Highest runup recorded for each tsunami: the top 12, coloured by cause. A runup is the height above sea level that the water reached on land. Hover a bar; click to see it on the map.
For earthquakes alone, the highest is 88 m, at Unga Island, AK after a quake in the Shumagin Islands of Alaska in 1788. In 2011, the Tohoku earthquake drove the water 39.3 m up the coast of Iwate; in 2004 it reached 50.9 m in Indonesia.
A big tsunami in deep water travels like a shallow wave over the whole ocean, at a speed set by the depth: about 710 km an hour over four kilometres of water. The records agree. For the 2,403 places more than 1,000 km from the source with a recorded travel time, the median speed works out at 680 km an hour (half lie between 600 and 730).
Distance against travel time
Every place with a recorded travel time and distance, coloured by water height. The dashed lines are steady speeds. Hover a dot. Loading the 32,628 places...
So the waves that crossed oceans came on a timetable. In 1946 an earthquake and landslide off Unimak Island in the Aleutians destroyed a lighthouse there and, 4.9 hours later and 3,810 km away, struck Hilo in Hawaii, where NCEI records 96 deaths. In 1960 the largest earthquake ever measured, magnitude 9.5 in southern Chile, sent water to Hilo after 14.8 hours (61 deaths there) and to Japan after 22.4 hours, about 17,050 km from Chile. In 2004 the Sumatra wave reached Kenya after 8.9 hours, Tanzania after 9.7 and South Africa after 11.1; Somalia, about 5,040 km away, saw water up to 9.5 m.
The farthest the water was measured
For each tsunami, the farthest place where a height above zero was recorded (sea waves only; see below for the eruptions' air waves). The top 10. Click a row to see it on the map.
The farthest measurement belongs to 2004: 0.2 m on a tide gauge at Baltra Island in the Galápagos, 19,266 km from Sumatra, 29 hours and 7 minutes after the earthquake, having crossed the Indian Ocean and much of the Pacific. Chilean earthquakes fill most of the rest of the list, because the Pacific carries their waves across to China, Taiwan and Japan, nearly half the world away. 75 tsunamis were measured more than 10,000 km from where they began.
Two eruptions moved the sea in a different way. NCEI marks 727 places where the water rose as the pressure wave from an eruption passed through the air: 643 for Hunga Tonga in 2022 and 84 for Krakatau in 1883. Because the air wave circled the globe, so did the disturbance: Hunga Tonga's water was recorded in 67 countries, the most of any event, as far as Tema in Ghana, 18,280 km away, and Krakatau's as far as Colon in Panama. I keep these places on the map but leave them out of the distance records above.
NCEI records a tsunami death toll for 244 of the 1,699 tsunamis (14%), and together they add up to 545,939 people. That is a floor, not a total: for 73 more NCEI gives only a rough order of magnitude, and for most old events nobody counted. The toll is dominated by a few days. The Indian Ocean tsunami of 26 December 2004 killed 227,899 people, 42% of every tsunami death in the database. The ten deadliest account for 78%.
The deadliest tsunamis
Deaths caused by the tsunami, as NCEI records them, for the 12 deadliest. Old figures are estimates. Hover a bar for the event's total, including the earthquake; click to see it on the map.
After 2004 come Lisbon in 1755 (50,000 tsunami deaths, from an earthquake that also shook the city down and set it on fire), Krakatau in 1883 (34,417), the Meiji Sanriku tsunami in Japan in 1896 (27,122), northern Chile in 1868 (25,000) and Tohoku in 2011 (18,423). Each figure is a town or a coastline emptied. Since 2000 alone, NCEI records 252,964 tsunami deaths.
NCEI also gives a second figure, the total deaths for the whole event, and it is easy to confuse the two. By that measure the deadliest entry is Haiti in 2010, with 316,000 deaths, but the tsunami that followed the earthquake killed 7. The Great Kanto earthquake of 1923 killed 105,385 people; its tsunami, 325. Everything on this page uses the tsunami's own toll.
The database starts with the eruption of Thera, dated to 1610 BC, but the early record is thin: only 4 tsunamis before AD 1 and 74 before 1500. Then it fills: 63 in the 1600s, 88 in the 1700s, 282 in the 1800s and 718 in the 1900s. 69% of the record is from 1900 or later.
Tsunamis recorded per century
Definite and probable tsunamis by century and cause, AD 1 to 2026 (the 4 earlier ones are left off). The 2000s are partial. Hover a bar.
This is not a rise in tsunamis. It is a rise in seeing them: writing spreading around the coasts, newspapers, then tide gauges in the 1800s, seismometers in the 1900s and deep-ocean pressure sensors since the 1990s, which catch waves of a few centimetres that nobody on shore would notice. Meteotsunamis show it clearly: 9 before 1950, 57 since 2000. Even in the instrument era the count creeps up, from 63 in 1900 to 1909 to 189 in 2010 to 2019. Since 1950 the database has averaged 11.5 tsunamis a year, and 1.4 a year that killed someone.
Counting the places where water was recorded, 140 countries and territories have been reached by at least one tsunami. Japan has been reached by the most: 395 different tsunamis, followed by the United States (264) and Indonesia (138). Japan also has the most measurements, 12,724, 38% of all places in the database, 6,326 of them from the survey after Tohoku in 2011.
The countries reached most often
Number of different tsunamis recorded in each country or territory (US and UK territories grouped). Hover a bar.
Data. NOAA NCEI's Global Historical Tsunami Database, fetched from the NCEI Hazard Service API in 2026: 3,138 source events and 35,129 runup records. NCEI rates each source from 4 (definite) to 3 (probable), 2 (questionable), 1 (very doubtful), 0 (a seiche or disturbance in an inland river or lake only) and minus 1 (an erroneous entry). I count only 3 and 4: 1,699 tsunamis, leaving out 1,439 entries. The cause codes are NCEI's: 1 earthquake, 2 questionable earthquake, 3 earthquake and landslide, 4 volcano and earthquake, 5 volcano, earthquake and landslide, 6 volcano, 7 volcano and landslide, 8 landslide, 9 meteorological, 10 explosion, 11 astronomical tide, 0 unknown. I checked them against events whose cause is not in doubt (the Halifax explosion is 10, the Vajont dam landslide 8, Krakatau 6) and grouped them as shown.
Places. A runup record is a place where NCEI records the water: a survey mark, a tide gauge or an eyewitness account. I keep the 33,532 records that belong to a definite or probable tsunami and are not flagged as doubtful (1,248 are). NCEI also flags places where the water was moved by an eruption's air wave or by the weather; I keep these, and leave the air wave places out of the distance records. 356 records come from deep-ocean pressure gauges far out at sea; they are left off the map and out of the country counts. Distance from the source and travel time are NCEI's. Oceans come from NCEI's region of the source; the Ring of Fire here means the regions around the rim of the Pacific (Japan, Kamchatka, Alaska, the Americas' Pacific coasts, the western Pacific islands, eastern Indonesia, New Zealand), not Hawaii; it leaves out the 63 tsunamis off Sumatra and Java, which NCEI places in the Indian Ocean.
Deaths. I use NCEI's count of deaths caused by the tsunami, never the total for the whole event, and I do not add up deaths over runup places, because one town can appear several times. Many historical figures are estimates, and some are disputed. The record before 1900 is very incomplete, and the counts per century mostly measure how much was written down.