Natural populations host a wealth of genetic variation in longevity and age-specific schedules of reproduction. This variation provides critical information for inferring the evolutionary origin of senescence. Patterns of mutational effects on age-specific fecundity and survival provide additional insight to distinguish alternative models of senescence. In this study, P-elements bearing the white minigene were inserted at random into a common genetic background, generating lines of D. melanogaster with single, stable transposon inserts. A series of 48 single-P-element lines revealed statistically significant heterogeneity in both longevity and fecundity. Longevity and early fecundity were only weakly positively correlated (r = 0.286, P = 0.0398). Both the pooled sample and 30 of the individual lines exhibited a leveling of age-specific mortality at advanced ages, in opposition to the classical demographic models. To the extent that these mutational effects are representative of naturally-occurring mutations in heterogeneous populations, this result presents a problem for the evolutionary theory of senescence. Natural selection is inefficient at removing deleterious mutations that are expressed only at late ages, and selection may not differentiate between mutations whose effects on longevity are post-reproductive. A leveling of the mortality rate would also be seen if mutations whose expression is delayed until very late simply do not occur. A simulation of mutation-selection balance among the 48 P-element tagged lines shows that the mean longevity declines monotonically with increasing mutation rate, consistent with the mutation-accumulation model.