Review of sea-level changes during the big five mass extinctions and several lesser extinction events reveals that the majority coincide with large eustatic inflexions. The degree of certainty with which these eustatic oscillations are known varies considerably. Thus, the late Ordovician and end Cretaceous extinctions are associated with unequivocal, major regressions demonstrated from numerous, widespread regions. In contrast, the multiple, high frequency sea-level changes reported for the Frasnian-Famennian crisis (based on the supposed depth-preferences of conodont taxa) have little support from sequence stratigraphic analyses, which reveals the interval to be one of highstand. The end Permian mass extinction has long been related to a severe, first order lowstand of sea level [Newell, N.D., 1967. Revolutions in the history of life. Geol. Soc. Am. Spec. Pap. 89, 63-91.] based primarily on the widespread absence of latest Permian ammonoid markers, but field evidence reveals that the interval coincides with a major transgression. Newell's hypothesis that marine extinctions are related to shelf habitat loss during severe regression remains tenable for the end Guadalupian and end Triassic extinction events but not for other crises. Rapid high amplitude regressive-transgressive couplets are the most frequently observed eustatic changes at times of mass extinction, with the majority of extinctions occurring during the transgressive pulse when anoxic bottom waters often became extensive. The ultimate cause of the sea-level changes is generally unclear. A glacioeustatic driving mechanism can only be convincingly demonstrated for the end Ordovician and end Devonian events. At other times, it is speculated that they may relate to the widespread regional doming (and subsequent collapse) caused by the impingement of superplumes (and ultimate eruption) on the base of the lithosphere.