The findings of the three experiments reported herein indicate that normal children can successfully teach social responses (i.e., delayed imitation, cooperative play, and verbalization of positive comments) to withdrawn mentally retarded peers. The effects of the intervention generalized across stimulus and response conditions, while the trained and generalized levels of responding were maintained after the end of the intervention. Moreover, the subjects developed social responding within their classrooms and play areas parallel to the intervention and continued to increase such responding after the interruption of the intervention. Direct edible reinforcement appeared to be necessary at least during the initial period of the intervention. Vicarious edible reinforcement seemed useful to prompt the appearance of responding. Vicarious social reinforcement was ineffective at the beginning of the intervention, but apparently acquired prompting power at a later stage of training. Generalization results indicated that the similarity between the response occasions used for training and those used for testing generalization played an important role. Yet, the extensiveness of training and the development of responding within the classrooms and play areas may also have had a relevant effect. The development of social responding within the classrooms and play areas appeared to be mainly the effect of new learning. This was perhaps due to vicarious and direct social reinforcement.DESCRIPTORS: social responses, generalization, maintenance, peer tutoring, normal children, retarded children One often-observed behavioral deficiency of severely mentally retarded children is their isolation from other children. Since many essential skills are learned and strengthened in the context of social interaction, promoting the interaction of retarded children with their peers has been the objective of several intervention programs. These programs were aimed at training imitation (Apolloni, Cooke, & Cooke, 1977;Guralnick, 1976;Paloutzian, Hasazi, Streifel, & Edgar, 1971;Peck, Apolloni, Cooke, & Raver, 1978) as well as at increasing positive physical contact, communication, and/or cooperation (Gable, Hendrickson, & Strain, 1978;Morris & Dolker, 1974;Petersen, Austin, & Lang, 1979;Strain, 1975Strain, , 1977Strain, Shores, & Kerr, 1976;Strain, Shores, & Timm, 1977;Strain & Timm, 1974;Whitman, Mercurio, & Caponigri, 1970;Young & Kerr, 1979).The results of these studies show that the social behavior of retarded children can be promoted. However, the findings concerning the maintenance of the acquired repertoire and/or its generalization across stimulus and response conditions are, overall, modest. Whitman et al. (1970) reported that during the training period the cooperative play of their subjects generalized to the classroom and involved peers and material not present in the training setting. 1982, 151,17-40 NUMBER I (SPRING 1982)