This paper addresses the question of whether our evolutionary history suggests that humans are likely to be individually selected selfish maximizers or group selected altruists. It surveys models from the literature of evolutionary biology in which groups are formed and dissolved and where the reproductive success of individuals is determined by their payoffs in a game played within groups. We show that if groups are formed randomly and reproductive success of group founders is determined by a multi-person prisoners’ dilemma game, then selfish behavior will prevail over maximization of group payoffs. However, interesting models can be found for which “group selection” sustains cooperative behavior. Forces that support cooperative behavior include assortative matching in groups, group longevity, and punishment-based group norms.