Selfish behavior can plague the formation of cooperative relationships and collective efforts. Understanding ways in which groups can overcome selfish motives and foster cooperation, therefore, becomes essential. Recent research reveals that reputation systems promote cooperation and deter antisocial behavior in groups. Little is known, however, about how and why people share reputational information. In this dissertation, I seek to establish the existence and dynamics of prosocial gossip, the sharing of negative evaluative information about a target in a way that protects others from antisocial or exploitative behavior. I present a model of prosocial gossip and the results of five studies testing the model's claims. Results of Studies 1-3 demonstrate that (a) individuals who observe an antisocial act experience negative affect and are compelled to share information about the antisocial actor with a potentially vulnerable person, (b) sharing such information reduces negative affect created by observing the antisocial behavior, (c) individuals possessing more prosocial orientations are the most motivated to engage in such gossip, even at a personal cost, and exhibit the greatest reduction in negative affect as a result. Taken together these results highlight the roles of prosocial motivations and negative affective reactions to injustice in maintaining reputational information sharing in groups. Studies 4-5 explore two ways in which prosocial gossip can effectively deter selfishness and promote cooperation. Study 4 reveals that prosocial gossip promotes cooperation by deterring selfish behavior, especially among those who are more egoistic and prone to exploit others. Study 5 demonstrates how prosocial gossip fosters cooperation by facilitating partner selection, guiding the recipients of the gossip in selecting who to interact with and who to ostracize. I conclude by discussing implications for reputational theories of the maintenance of cooperation in human groups and laying out possibilities for future research.