The causes of mass extinctions and the nature of biological selectivity at extinction events are central questions in palaeobiology. It has long been recognized, however, that the amount of sedimentary rock available for sampling may bias perceptions of biodiversity and estimates of taxonomic rates of evolution. This problem has been particularly noted with respect to the principal mass extinctions. Here we use a new compilation of the amount of exposed marine sedimentary rock to predict how the observed fossil record of extinction would appear if the time series of true extinction rates were in fact smooth. Many features of the highly variable record of apparent extinction rates within marine animals can be predicted on the basis of temporal variation in the amount of exposed rock. Although this result is consistent with the possibility that a common geological cause determines both true extinction rates and the amount of exposed rock, it also supports the hypothesis that much of the observed short-term volatility in extinction rates is an artefact of variability in the stratigraphic record.