In this paper we develop predictions from models of life-long demographic heterogeneity. These predictions are then compared to observations of mortality in large laboratory populations of Drosophila melanogaster. We find that the demographic heterogeneity models either require levels of variation that far exceed what would be considered biologically plausible, or they predict a much larger number of very old individuals than we actually observe. We conclude that the demographic heterogeneity models are not reasonable explanations of demographic patterns and are weakly motivated biological models.