Mammalian play is believed to improve motor skills as well as facilitate the development of social relationships. Given the marked sexual dimorphism in gorilla body size and the role assumed by the male in protecting the group from conspecifics and predators, the motor-training hypothesis of play predicts that male infants should exhibit higher frequencies of social play than female infants, and that males should prefer to play with other males. Given that adult female gorillas are strongly attracted to adult breeding males and form only weak social bonds with unrelated adult females, the social-relationship hypothesis of play predicts that female infants should prefer to play with males. These hypotheses were tested in a 22-month study of 12 gorilla infants, aged between 0–5 years, living in three zoological parks in Chicago and Atlanta. Consistent with the hypotheses, male infants played more than female infants did, and both male and female infants preferred to play with males rather than with females. These findings suggest that sex differences in play in the great apes and other primates can be predicted by the characteristics of adult behavior and social structure above and beyond the patterns of sex-biased dispersal or coalition formation with same-sex kin.