This paper reports the results of five experiments on the disruption and recovery of social exchange and cooperation, and then uses these results as the empirical base for theory-building. The experiments investigate the effects of changes in social and individual reinforcement contingencies on the disruption of interaction, and the effects of structural variations in the social contingencies on recovery after disruption. The results suggest that contingency changes are disruptive even if social interaction remains the more rewarding response. Explanations of the effects of different contingency changes are offered. Once interaction is disrupted, the chances of recovery are influenced by (1) the actual reinforcement experienced during disruption for social and nonsocial responses, and (2) the ease of discriminating the deviant partner's recovery attempts. The linkages between these variables and the structural contingencies are discussed.