Animal societies provide perfect conditions for the spread of infections and are therefore expected to employ mechanisms that reduce the probability of transmitting pathogens to group members [1-4]. Death in nature rarely results from old age but commonly results from diseases. Leaving one's group to die in seclusion might be an efficient way of minimizing the risk of infecting kin. Anecdotal observations of moribund individuals deserting from their groups exist for several species, including humans (e.g., ), but have rarely been substantiated by quantitative analysis. Furthermore, to confirm that dying in solitude has evolved because of its altruistic benefits requires refuting the alternative explanation of pathogen manipulation of host behavior. Here we show that workers of the ant Temnothorax unifasciatus dying from fungal infection, uninfected workers whose life expectancy was reduced by exposure to 95% CO(2)[6, 7], and workers dying spontaneously in observation colonies exhibited the same suite of behavior of isolating themselves from their nestmates days or hours before death. Actively leaving the nest and breaking off all social interactions thus occurred regardless of whether individuals were infected or not. Social withdrawal might be a commonly overlooked altruistic trait serving the inclusive fitness interests of dying individuals in social animals.