Two models of faunal turnover patterns, one with constant turnover and another with climatically induced turnover pulses, were tested against the empirical fossil data of first and last appearances of large mammals from the late Pliocene and Pleistocene of East Africa. Computer simulations of each model were generated by first creating change in hypothetical faunal communities and then sampling the evolving communities in a manner scaled to the specific contingencies of the East African fossil record. Predictions of the two turnover models were compared with the empirical data. Neither model yielded predictions that deviated significantly from the observed patterns of first and last appearances of species, and both models produced extremely similar results. The implication is that the fossil data of East Africa are not refined enough to detect variations in the pace of turnover; the patterns of first- and last-appearance frequencies are determined more by the contingencies of the fossil record than by the underlying evolutionary and migrational patterns. Whereas these results undermine the primary basis of empirical support for the turnover-pulse hypothesis, they do not imply that other models are more likely. However, the simulation results were highly suggestive of significant reduction in species biodiversity of large mammals during the past 2 Myr.