Occurrence of major earthquakes is as stochastic as smaller ones

Seismic hazard estimates rely on interevent time distributions between earthquakes of a given magnitude. In the Himalaya, recurrence intervals are usually modeled as cyclic or quasiperiodic, whereas globally, they range from periodic and clustered to random. Statistical analyses of a 6000-year lake-sediment seismic record, calibrated against regional instrumental data, worldwide paleoseismic records, and synthetic seismic catalogs, demonstrate that time intervals between large earthquakes (M ≥ 6.5, based on shaking intensity thresholds calibrated locally) robustly follow a Poisson distribution. Second-order fluctuations indicate event clustering. These observations contradict periodic or quasiperiodic recurrence models. Comparisons with paleoseismic data from other tectonic settings and realistic synthetic catalogs confirm the robustness and broad applicability of these findings. Thus, major earthquakes appear as stochastic as smaller ones, challenging recurrence models derived from limited datasets and substantially increasing seismic hazard estimates.