Most earthquakes that people actually feel last somewhere between a few seconds and a couple of minutes, but the rupture process itself can range from a fraction of a second for a tiny tremor to roughly eight minutes for the largest events ever recorded. The 2004 Sumatra-Andaman earthquake, one of the most powerful in modern history, ruptured for about 500 seconds along a fault stretching over 1,200 kilometers. What controls that duration is surprisingly straightforward in principle: how long a crack in the Earth takes to unzip depends on how long the fault is and how fast the rupture travels along it. In practice, though, a tangle of geological factors can push those two variables in unexpected directions.
Fault Length and Rupture Speed Set the Clock
An earthquake’s duration is, at its core, the time it takes for a fracture to propagate from where it starts to where it stops. Think of it like tearing a piece of fabric: a short rip finishes quickly, a long rip takes longer, and how fast you pull determines the pace. The two main ingredients are therefore the length of the fault that breaks and the speed at which the rupture front moves.
For the 2004 Sumatra-Andaman earthquake, researchers estimated a rupture length of about 1,220 kilometers, a duration of about 500 seconds, and a mean rupture velocity of 2.4 kilometers per second.1Geophysical Research Letters. Rupture length and duration of the 2004 Aceh‐Sumatra earthquake from the phases of the Earth’s gravest free oscillations That works out to a little over eight minutes of continuous fault rupture, the longest well-documented rupture duration for a single earthquake. A moderate magnitude-5 earthquake, by contrast, typically breaks a fault segment only a few kilometers long and finishes in a second or two.
Fault scaling research shows that the relationship between magnitude and rupture length is not a simple straight line. For earthquakes below about magnitude 5, the fault breaks in all three dimensions more or less proportionally. Above magnitude 5, the fault’s width becomes limited by the thickness of the brittle crust, so the rupture grows mostly in length. For very large strike-slip earthquakes above about magnitude 7.2, this width limitation becomes even more constraining.2Bulletin of the Seismological Society of America. Earthquake Fault Scaling: Self-Consistent Relating of Rupture Length, Width, Average Displacement, and Moment Release The upshot is that bigger earthquakes don’t just rupture longer faults; they do so in a way that makes duration climb steeply with magnitude once you pass a certain size threshold.
The material the fault cuts through also matters. Earthquakes along oceanic transform faults tend to have longer ruptures than continental earthquakes of the same magnitude, likely because the hydrated oceanic crust is less rigid, allowing fractures to stretch further for a given amount of energy release.3Geophysical Research Letters. Relationship Between Rupture Length and Magnitude of Oceanic Transform Fault Earthquakes A longer rupture at a similar speed means a longer-duration earthquake.
When Ruptures Outrun Their Own Waves
Rupture speed is not fixed. Most earthquake ruptures travel at roughly 70 to 90 percent of the local shear-wave speed, which in crustal rock is typically around 3 to 4 kilometers per second. But under certain conditions a rupture can accelerate past the shear-wave speed into what seismologists call the “supershear” regime, analogous to an aircraft breaking the sound barrier. A supershear rupture compresses the earthquake’s energy release into a shorter time window and concentrates shaking in a narrow zone, much like a sonic boom.
This phenomenon was dramatically documented during a magnitude 6.7 aftershock of the 2013 Sea of Okhotsk earthquake, where the rupture tore downward along a steeply dipping fault at an average speed of about 8 kilometers per second, far above the shear-wave speed at that depth.4PubMed. Supershear rupture in a M(w) 6.7 aftershock of the 2013 Sea of Okhotsk earthquake Because the rupture moved so fast, it finished sooner than a comparable earthquake rupturing at normal speed would have.
The friction properties of the fault surface play a key role in whether supershear happens. Research on rate-and-state friction modeling shows that certain friction parameters primarily control how fast the rupture propagates, while others govern how much stress is released. There is often an optimal combination that maximizes rupture velocity, meaning supershear is not random but depends on specific physical conditions along the fault.5Geophysical Journal International. Diversity and transition of rupture styles governed by rate-and-state friction
Tsunami Earthquakes and Unusually Slow Ruptures
On the opposite end of the speed spectrum are “tsunami earthquakes,” a term for events whose ruptures crawl along so slowly that they generate disproportionately large ocean waves relative to how strong the shaking feels on land. These are among the most deceptive natural hazards: the ground shaking might seem only moderate, leading people near the coast to underestimate the incoming wave.
The 1947 Hikurangi margin earthquake off New Zealand is a well-studied example. Modeling of its seismic records requires an unusually low rupture velocity of just 150 to 300 meters per second, compared to the several-kilometers-per-second speeds seen in ordinary earthquakes. At those speeds, tsunami run-up heights more than doubled compared to what an instantaneous rupture of the same size would produce. The slow rupture was likely connected to a subducted seamount that created complex, hesitant fracture propagation.6Earth and Planetary Science Letters. Hikurangi margin tsunami earthquake generated by slow seismic rupture over a subducted seamount
The 1992 Nicaragua earthquake showed a similar pattern. It was a slow thrust earthquake where the rupture propagated through soft subducted sediments on the plate interface, making the entire process slower than a typical subduction-zone event. Because the slip extended all the way to the ocean floor without the sediment buffer that usually dampens displacement, it excited large tsunamis despite relatively modest ground shaking on land.7Nature. The 1992 Nicaragua earthquake: a slow tsunami earthquake associated with subducted sediments
Dynamic rupture simulations have explored what causes these slow speeds. When the fracture energy required to break the rock is tripled (by increasing the critical distance over which frictional weakening occurs), the modeled rupture velocity drops dramatically and no supershear rupture occurs at all, reproducing the behavior of a tsunami earthquake.8Frontiers in Earth Science. 3D Linked Subduction, Dynamic Rupture, Tsunami, and Inundation Modeling: Dynamic Effects of Supershear and Tsunami Earthquakes, Hypocenter Location, and Shallow Fault Slip In short, when the fault is harder to break, the rupture takes its time, and the earthquake lasts longer for its magnitude than you would expect.
Slow Slip Events That Last Hours to Years
If the spectrum of earthquake duration already seems wide, it gets much wider when you include slow slip events. These are episodes where tectonic plates slide past each other in a way that releases seismic energy far too slowly to generate the kind of shaking you would feel. They were only discovered in the late 1990s and early 2000s when GPS networks became sensitive enough to detect the creeping motion.
Slow slip events span an enormous range of timescales. Some last just hours: researchers in Japan’s Tokai region identified temporary slip-speed increases lasting one to three hours during short-term slow slip events, detected by borehole and laser strainmeters between 2016 and 2022. These brief pulses were accompanied by tectonic tremor, a faint seismic hum that acts as an acoustic fingerprint of slow fault motion.9Earth, Planets and Space. Temporary slip speed increases during short-term slow slip events with durations of one to three hours
At the other extreme, a slow slip event on the south-central Alaska megathrust between 2009 and 2013 lasted about five years and released energy equivalent to roughly a magnitude-7.6 earthquake. Within that multi-year episode, weeks-long bursts of faster slip accompanied tectonic tremor. During one such burst in September 2010, the slip migrated along the fault at about 8 kilometers per day, with slip rates of around 3 millimeters per day. GPS stations recorded velocities three to six times higher during tremor bursts than in the quiet intervals between them.10Journal of Geophysical Research: Solid Earth. Weeks‐Long and Years‐Long Slow Slip and Tectonic Tremor Episodes on the South Central Alaska Megathrust
Slow slip events are not directly dangerous in the way a sudden earthquake is, but they matter for earthquake duration science because they reveal that the boundary between “earthquake” and “not earthquake” is blurry. The same fault interface can host a catastrophic seconds-long rupture, a moderate months-long creep episode, and everything in between. Understanding where a fault sits on that spectrum is one of the harder problems in seismology.
How Depth Changes the Picture
Earthquakes happen at depths ranging from the surface to about 700 kilometers, and depth has a measurable effect on rupture characteristics. Deep-focus earthquakes, those occurring in the mantle transition zone hundreds of kilometers down, face very different physical conditions than shallow crustal events. The stress drop in deep earthquakes, a measure of how much stored stress is released when the fault slips, is about ten times higher than for shallow earthquakes of the same magnitude.11Geophysical Research Letters. Stress Drop Variation of Deep‐Focus Earthquakes Based on Empirical Green’s Functions
A higher stress drop means the fault releases its energy more abruptly. For a given magnitude, a high-stress-drop earthquake will typically have a shorter rupture area and a more compact, intense burst of energy compared to a low-stress-drop event that stretches the energy release over a bigger patch of fault. This is one reason deep earthquakes can feel sharp and punchy at the surface while shallow ones of the same magnitude sometimes produce longer, rolling shaking. The surrounding rock at great depth is also hotter, under immense pressure, and behaves differently, so the mechanics of how faults break at those depths are still debated.
Cascading Ruptures and Earthquake Doublets
Not every earthquake is a single clean fracture. Some of the most destructive events involve rupture cascading across multiple fault segments, and this cascading process can significantly extend the total duration. Whether a rupture jumps from one fault to another depends on how the faults are oriented relative to the regional stress field, how close together they are, and how much energy is available at the rupture front.
Three-dimensional simulations of complex fracture networks show that cascading ruptures are dynamically feasible when fracture energy scales with fault size, when faults are favorably aligned with the ambient stress, and when fractures are close enough to interact. An interesting finding is that cascading rupture within a surrounding damage zone can actually discourage rupture on the main fault, meaning the earthquake’s energy gets distributed across many small fractures rather than channeled into one large break.12Journal of Geophysical Research: Solid Earth. Rupture Dynamics of Cascading Earthquakes in a Multiscale Fracture Network
Earthquake doublets are an extreme version of this. The devastating 2023 Türkiye earthquakes are a textbook case: two great earthquakes struck within nine hours of each other on different fault systems. The first event involved supershear rupture on some segments and slower rupture on others, with the geometry and pre-stress of the multiple fault segments controlling the rupture behavior and shaking intensity.13PubMed. Supershear triggering and cascading fault ruptures of the 2023 Kahramanmaraş, Türkiye, earthquake doublet The second event was triggered when the first earthquake changed the stress field on a neighboring fault, with normal stress changes exceeding 0.8 megapascals that effectively unclamped the second fault.14Communications Earth & Environment. Distinct triggering mechanisms of the 2023 Türkiye earthquake doublet
An earlier example from the central Kuril Islands illustrates how different the two halves of a doublet can be. In 2006, a magnitude-8.3 underthrusting earthquake ruptured the shallow plate boundary. Within minutes, extensional earthquakes began in the outer rise seaward of the trench, and two months later a magnitude-8.1 normal-fault event broke through the upper Pacific plate in one of the largest shallow extensional earthquakes ever recorded.15Nature. A great earthquake doublet and seismic stress transfer cycle in the central Kuril islands If you experienced both events, the effective duration of major earthquake shaking in that region stretched from one episode lasting minutes to a sequence playing out over weeks.
Why Shaking Can Outlast the Earthquake Itself
There is an important distinction between how long the fault actually ruptures and how long you feel shaking at the surface. The rupture duration is a property of the earthquake source, but the shaking duration at any given location also depends on what the seismic waves pass through on their way to you.
Sedimentary basins are notorious for amplifying and prolonging shaking. Simulations of a potential southern San Andreas fault earthquake show that the chain of sedimentary basins between San Bernardino and downtown Los Angeles acts as a waveguide, channeling seismic waves along the southern edge of the San Gabriel Mountains. In certain rupture scenarios, this produces unusually high long-period ground motions across much of greater Los Angeles, with intense localized variations caused by changes in the waveguide’s shape.16Geophysical Research Letters. Strong shaking in Los Angeles expected from southern San Andreas earthquake In these scenarios, the shaking people feel can last considerably longer than the rupture itself, because the basin keeps reverberating after the source has stopped.
This basin effect is confirmed by physics-based 3D ground motion modeling, which shows that simplified one-dimensional site analysis can actually overestimate shaking duration in sedimentary basins because of trapped waves that bounce around within the basin structure. The real three-dimensional response involves waves scattering and losing energy as they interact with the basin’s edges and bottom, but the net effect is still a prolonged shaking experience compared to sites on solid rock.17Earthquake Engineering & Structural Dynamics. Applicability of 1D site response analysis to shallow sedimentary basins: A critical evaluation through physics‐based 3D ground motion simulations
For anyone living in a basin city like Los Angeles, Mexico City, or Kathmandu, the practical takeaway is that a large earthquake on a nearby fault will shake for longer than the same earthquake would on bedrock. Building codes in basin areas attempt to account for this by requiring structures to handle longer-duration shaking, but the science of predicting exactly how long is still catching up to the complexity of the geology.
Fluid Injection and Induced Earthquakes
Human activities like wastewater injection, hydraulic fracturing, and geothermal energy extraction can trigger earthquakes, and these induced events sometimes have rupture characteristics that differ from natural ones. Research on fluid-injection-induced earthquakes has identified a category of events with hybrid-frequency waveforms that sit between typical fast earthquakes and slow aseismic slip. These events show much lower stress drops (averaging about 0.29 megapascals compared to 4.86 megapascals for typical induced events in the same area) or much lower rupture speeds (averaging about 1.1 kilometers per second compared to 2.3 kilometers per second for ordinary induced events), or some combination of both.18Nature Communications. Fluid-injection-induced earthquakes characterized by hybrid-frequency waveforms manifest the transition from aseismic to seismic slip
These hybrid events appear to represent a transitional state between slow aseismic slip and full seismic rupture, suggesting that the fluid pressure changes introduced by injection push faults through a range of failure modes. For the question of duration, this means induced earthquakes near injection sites can behave as if the fault is struggling to decide whether to slip slowly or break suddenly, with the rupture either crawling along at reduced speed or spreading over a larger area at low stress drop, both of which elongate the source process relative to a normal earthquake of the same magnitude.
How Scientists Measure Rupture Duration
Knowing how long an earthquake’s rupture lasted is not as simple as starting and stopping a timer. The rupture happens kilometers underground, and what seismometers record is the combined effect of the source, the path the waves traveled, and the local ground conditions. Separating those three contributions is a core challenge.
One of the most important techniques developed in recent years is back-projection imaging, which emerged after the 2004 Sumatra-Andaman earthquake demonstrated how inadequate existing tools were for tracking the rupture of enormous events. The method uses large arrays of seismometers at great distances from the earthquake to trace where the seismic energy came from at each moment, effectively making a movie of the rupture as it unfolded.19Annual Review of Earth and Planetary Sciences. Back-Projection Imaging of Earthquakes Its strength is that it makes few assumptions about how the fault is shaped or how the rupture moves, letting the data reveal complex, irregular propagation patterns that simpler models would miss.
A decade of applying multi-array back-projection to large earthquakes has built a growing catalog of detailed rupture histories. Combining data from multiple seismometer arrays and processing both direct and reflected seismic wave phases has steadily improved the resolution of these rupture images.20Journal of Geophysical Research: Solid Earth. A Decade of Short‐Period Earthquake Rupture Histories From Multi‐Array Back‐Projection However, the reliability of back-projection images has not been as thoroughly tested with synthetic benchmarks as researchers would like, and recent work on validating the method against known artificial ruptures suggests that some features in published back-projection images may be artifacts rather than real rupture properties.21Geophysical Journal International. Synthetic tests of reliability of back-projection images of earthquake ruptures
For earthquake early warning, the challenge is even steeper: you need to estimate how big an earthquake will be while it is still happening. Newer approaches that combine seismic and geodetic data (essentially pairing traditional seismometers with GPS-like instruments that measure ground displacement directly) can produce magnitude estimates within about two to three minutes of earthquake initiation for events of magnitude 7 and above, with an accuracy of about 0.2 magnitude units.22Journal of Geophysical Research: Solid Earth. Rapid Earthquake Magnitude Estimation for Local Early Warning Systems Using Seismogeodetic That speed matters because for a very large earthquake whose rupture might last several minutes, the warning system needs to work faster than the earthquake itself unfolds. Getting the duration and magnitude right in real time remains one of the field’s hardest open problems.