Eusocial insects ants, bees, wasps and termites are being recognized as model organisms to unravel the evolutionary paradox of aging for two reasons: (1) queens (and kings, in termites) of social insects outlive similar sized solitary insects by up to several orders of magnitude; (2) all eusocial taxa show a divergence of long queen and shorter worker lifespans, despite their shared genomes and even under risk-free laboratory environments. Traditionally, these observations have been explained by invoking classical evolutionary aging theory: well-protected inside their nests, queens are much less exposed to external hazards than foraging workers, and this provides natural selection the opportunity to favor queens that perform well at advanced ages. Although quite plausible, these verbal arguments have not been backed up by mathematical analysis. Here, for the first time, we provide quantitative models for the evolution of caste specific aging patterns. We show that caste-specific mortality risks are in general neither sufficient nor necessary to explain the evolutionary divergence in lifespan between queens and workers and the extraordinary queen lifespans. Reproductive monopolization and the delayed production of sexual offspring in highly social colonies lead natural selection to inherently favor queens that live much longer than workers, even when exposed to the same external hazards. Factors that reduce a colony's reproductive skew, such as polygyny and worker reproduction, tend to reduce the evolutionary divergence in lifespan between queens and workers. Caste-specific extrinsic hazards also affect lifespan divergence but to a much smaller extent than reproductive monopolization.