Late-life plateaus in age-specific mortality have been an evolutionary and biodemographic puzzle for decades. Although classic theory on the evolution of senescence predicts late-life walls of death, observations in experimental organisms document the opposite trend: a slowing in the rate of increase of mortality at advanced ages. Here, I analyze published life-history data on individual Drosophila melanogaster females and argue for a fundamental change in our understanding of mortality in this important model system. Mortality plateaus are not, as widely assumed, exclusive to late life, and are not explained by population heterogeneity-they are intimately connected to individual fecundity. Female flies begin adult life in the working stage, a period of active oviposition and low but accelerating mortality. Later they transition to the retired stage, a terminal period characterized by limited fecundity and relatively constant mortality. Because ages of transition differ between flies, age-synchronized cohorts contain a mix of working and retired flies. Early- and mid-life plateaus are obscured by the presence of working flies, but can be detected when cohorts are stratified by retirement status. Stage-specificity may be an important component of Drosophila life-history evolution.