Using older and in part flawed data, Ruff (1989) suggested that thick sediment entering the subduction zone (SZ) smooths and strengthens the trench-parallel distribution of interplate coupling. This circumstance was conjectured to favor rupture continuation and the generation of high-magnitude (≥Mw8.0) interplate thrust (IPT) earthquakes. Using larger and more accurate compilations of sediment thickness and instrumental (1899 to January 2013) and pre-instrumental era (1700–1898) IPTs (n = 176 and 12, respectively), we tested if a compelling relation existed between where IPT earthquakes ≥Mw7.5 occurred and where thick (≥1.0 km) versus thin (≤1.0 km) sedimentary sections entered the SZ.

Based on the new compilations, a statistically supported statement (see Summary and Conclusions) can be made that high-magnitude earthquakes are most prone to nucleate at well-sedimented SZs. For example, despite the 7500 km shorter global length of thick-sediment trenches, they account for ∼53% of instrumental era IPTs ≥Mw8.0, ∼75% ≥Mw8.5, and 100% ≥Mw9.1. No megathrusts >Mw9.0 ruptured at thin-sediment trenches, whereas three occurred at thick-sediment trenches (1960 Chile Mw9.5, 1964 Alaska Mw9.2, and 2004 Sumatra Mw9.2).

However, large Mw8.0–9.0 IPTs commonly (n = 23) nucleated at thin-sediment trenches. These earthquakes are associated with the subduction of low-relief ocean floor and where the debris of subduction erosion thickens the plate-separating subduction channel. The combination of low bathymetric relief and subduction erosion is inferred to also produce a smooth trench-parallel distribution of coupling posited to favor the characteristic lengthy rupturing of high-magnitude IPT earthquakes. In these areas subduction of a weak sedimentary sequence further enables rupture continuation.