It is widely believed (following the 1957 hypothesis of G. C. Williams) that greater rates of "extrinsic" (age- and condition-independent) mortality favor more rapid senescence. However, a recent analysis of mammalian life tables failed to find a significant correlation between minimum adult mortality rate and the rate of senescence. This article presents a simple theoretical analysis of how extrinsic mortality should affect the rate of senescence (i.e., the rate at which probability of mortality increases with age) under different evolutionary and population dynamical assumptions. If population dynamics are density independent, extrinsic mortality should not alter the senescence rate favored by natural selection. If population growth is density dependent and populations are stable, the effect of extrinsic mortality depends on the age specificity of the density dependence and on whether survival or reproduction (or both) are functions of density. It is possible that higher extrinsic mortality will increase the rate of senescence at all ages, decrease the rate at all ages, or increase it at some ages while decreasing it at others. Williams's hypothesis is most likely to be supported when density dependence acts primarily on fertility and does not differentially decrease the fertilities of older individuals. Patterns contrary to Williams's prediction are possible when density dependence acts primarily on the survival or fertility of later ages or when most variation in mortality rates is due to variation in nonextrinsic mortality.