The Largest Earthquakes in Recorded History

Instrumental seismology dates back only to about 1900, and reliable global coverage to the 1960s. Every entry below is therefore from a very short slice of geological time, and earthquakes larger than any of these have certainly occurred before records began.

The ten largest instrumentally recorded

1. Valdivia, Chile — 22 May 1960 — M9.5. The largest earthquake ever recorded. Roughly 1,000 km of the boundary between the Nazca and South American plates ruptured. Shaking lasted around ten minutes. The tsunami crossed the Pacific, killing people in Hawaii, Japan and the Philippines. Coastal land subsided by up to two metres, permanently flooding farmland.

2. Prince William Sound, Alaska — 27 March 1964 — M9.2. Known as the Good Friday earthquake. Shaking continued for four and a half minutes. Extensive liquefaction and submarine landslides destroyed parts of Anchorage and Valdez. Vertical land movement exceeded 11 metres in places. It prompted major advances in understanding of subduction and directly motivated the Pacific tsunami warning infrastructure.

3. Sumatra–Andaman, Indonesia — 26 December 2004 — M9.1–9.3. Approximately 1,300 km of the Sunda megathrust ruptured over about ten minutes. The resulting tsunami killed roughly 228,000 people across fourteen countries, the deadliest tsunami in recorded history. Its most consequential legacy was the establishment of an Indian Ocean warning system, which had not previously existed.

4. Tōhoku, Japan — 11 March 2011 — M9.0–9.1. Struck a country with the world's most advanced earthquake preparedness. Building performance was excellent and the early warning system worked. But the tsunami exceeded design assumptions, overtopping sea walls, killing around 18,500 people and causing the Fukushima Daiichi nuclear accident. A case study in how tsunami hazard can be underestimated even where seismic hazard is well managed.

5. Kamchatka, Russia — 4 November 1952 — M9.0. Generated a Pacific-wide tsunami that caused significant damage in Hawaii. Sparsely populated, so casualties were low.

6. Maule, Chile — 27 February 2010 — M8.8. Modern building codes, developed in the wake of 1960, performed well. Around 500 deaths, a remarkably low toll for an event of this size, most caused by the tsunami rather than shaking.

7. Ecuador–Colombia — 31 January 1906 — M8.8. Ruptured roughly 500 km of coastline and generated a destructive regional tsunami.

8. Rat Islands, Alaska — 4 February 1965 — M8.7. Remote, with minimal casualties.

9. Assam–Tibet — 15 August 1950 — M8.6. Unusual for this list in being a continental collision earthquake rather than an oceanic subduction event, reflecting the India–Eurasia convergence that built the Himalaya. Caused massive landslides and river blockages.

10. Northern Sumatra — 26 December 2004 (aftershock sequence) and 28 March 2005 — M8.6. The March 2005 Nias earthquake followed the great 2004 event on an adjacent segment.

What they have in common

Nine of the ten were subduction thrust earthquakes. As explained in the plate tectonics guide, this is not coincidence — it is geometry. Magnitude scales with rupture area, and only a shallow-dipping plate interface offers the combination of great length and great width required.

All were shallow, which is a precondition for a large interface rupture and also for tsunami generation. All but one occurred on the Pacific Ring of Fire.

And in most cases, the tsunami killed more people than the shaking. This is the consistent lesson of the great earthquakes: for coastal populations near a subduction zone, water is the primary threat.

The deadliest earthquakes are a different list

Ranking by casualties produces almost no overlap.

The 1556 Shaanxi earthquake in China is estimated to have killed around 830,000 people, largely because the population lived in cave dwellings carved into loess cliffs, which collapsed catastrophically. Its magnitude is estimated at about 8.

The 2010 Haiti earthquake was magnitude 7.0 — over 300 times less energetic than the 1960 Chile event — yet killed well over 100,000 people. It was shallow, close to Port-au-Prince, and struck a city of largely unreinforced concrete construction with no seismic code enforcement.

The 1976 Tangshan earthquake in China, magnitude 7.6, killed at least 242,000.

Compare Chile in 2010: magnitude 8.8, roughly 500 deaths. The difference is not luck. It is building codes, enforcement, and a national memory of 1960.

The variables that determine casualties are, roughly in order: construction quality, depth, proximity to population, time of day, and tsunami exposure. Magnitude appears well down that list.

Before instrumental records

Historical and geological evidence extends the record much further back, with correspondingly larger uncertainty.

The 1700 Cascadia earthquake off the Pacific Northwest of North America is estimated at magnitude 8.7–9.2. Its date and approximate size were established by an unusually elegant chain of evidence: Japanese records describe an "orphan tsunami" arriving on 27 January 1700 with no local earthquake, and tree-ring dating of drowned coastal forests in Washington and Oregon matches the same winter. Native American oral traditions preserved accounts of the event.

The 1755 Lisbon earthquake, estimated around magnitude 8.5–9.0, destroyed the city through shaking, fire and tsunami, killing tens of thousands. It had a profound intellectual impact across Europe, prompting some of the first systematic attempts to study earthquakes as natural rather than divine phenomena — and is often cited as the birth of seismology as a discipline.