Bats live longer than similar-sized mammals, but the number of lineages that have independently evolved extreme longevity has not previously been determined. Here we reconstruct the evolution of size-corrected longevity on a recent molecular phylogeny and find that at least four lineages of bats have lifespans more than fourfold those of similar-sized placental mammals, with the ancestral bat projected to live 2.6 times as long. We then evaluate a series of phylogenetic generalized least-squares models containing up to nine variables hypothesized to influence extrinsic mortality. These analyses reveal that body mass and hibernation predict longevity. Among hibernators, longevity is predicted by the absolute value of the median latitude of the species range and cave use, while cave use and lack of sexual dimorphism predict longevity among non-hibernators. The importance of torpor in extending lifespan is further supported by the one lineage with extreme longevity that does not hibernate but exhibits flexible thermoregulation, the common vampire bat. We propose several potential mechanisms that may enable bats to live so long, and suggest that the ability to tolerate a wide range of body temperatures could be important for surviving viral or other pathogen infections.