A series of four studies (employing combinations of multiple baseline and reversal designs) examined the effects of both antecedent and consequent events on the social interactions of four primary grade withdrawn children. These four experiments were conducted during the first year of pilot-testing a program aimed at increasing the social interaction skills of withdrawn children in the regular primary grades. The experiments were all short term interventions designed to evaluate the application of different treatment procedures. The specific procedures tested were: (1) joint task (assignment of the subject and a peer to work at a specific task requiring alternating verbal interaction); (2) individually contingent teacher praise; and (3) three forms of contingent token reinforcement: individually contingent praise plus points with individual backups; individually contingent praise plus points with a backup shared by the entire group; and a crossover contingency in which the subject earned points for group initiations, the group earned points for subject initiations, and the backups were delivered only after both the subject and the group met a specific criterion. The four studies, in various combinations, compared: (1) an individual praise contingency to joint task; (2) the three forms of contingent token reinforcement; and (3) the effects of the individual reinforcement group backup procedure in an academic setting to one implemented during recess. Results suggest that the joint task procedure is most effective in the classroom setting and that the individual reinforcement contingency with a group backup is most effective during recess. (Author/BD)