Diet effects on age-dependent mortality patterns are well documented in a large number of animal species, but studies that look at the effects of nutrient availability on late-life mortality plateaus are lacking. Here, we focus on the effect of dietary protein content (low, intermediate, and high) on mortality trajectories in late life in the fruit fly Drosophila melanogaster. According to the two theories that are mainly implicated in explaining the deceleration of mortality rate in late life (the heterogeneity/frailty theory and the Hamiltonian theory), we predict, in general, the occurrence of late-life mortality deceleration under most circumstances, independent of sex and dietary regime. However, the heterogeneity theory of late life is more flexible in allowing no mortality deceleration to occur under certain circumstances compared with the Hamiltonian theory. We applied a novel statistical approach based on Bayesian inference of age-specific mortality rates and found a deceleration of late-life mortality rates on all diets in males but only on the intermediate (standard) diet in females. The difference in mortality rate deceleration between males and females on extreme diets suggests that the existence of mortality plateaus in late life is sex and diet dependent and, therefore, not a universal characteristic of large enough cohorts.