The demonstration of life span plasticity in natural populations would provide a powerful test of evolutionary theories of senescence. Plastic senescence is not easily explained by mutation accumulation or antagonistic pleiotropy but is a corollary of the disposable soma theory. The life span differences among castes of the eusocial Hymenoptera are potentially some of the most striking and extreme examples of life span plasticity. Although these differences are often assumed to be plastic, this has never been demonstrated conclusively because differences in life span may be caused by the proximate effects of different levels of environmental hazard experienced by castes. Here age-dependent and age-independent components of instantaneous mortality rates of the honey bee (Apis mellifera) were estimated from published life tables for natural and seminatural populations to determine whether differences in life span between queens and workers and between different types of workers are indeed plastic. These differences in life span were found to be due to differences in the rate of actuarial senescence, which correlate positively with the rate of extrinsic mortality, in accordance with the central prediction of evolutionary theories of senescence. Although all three evolutionary theories of senescence could in principle explain such plastic senescence, given differential gene expression between castes or life stages, only the disposable soma theory adequately explains the adaptive regulation of somatic maintenance in response to different environmental conditions that appears to underlie life span plasticity.