Play skills were taught to eight profoundly mentally retarded adults in two interrelated experiments. In Experiment 1, a multiple baseline across subjects design was used to assess the efficacy of verbal and physical prompts on independent play. In Experiment 2, the same subjects and experimental procedures were used to develop social play. Verbal prompting and graduated physical guidance procedures were found to be effective in substantially increasing independent play in Experiment 1 and social play in Experiment 2. Positive changes were also observed in collateral behaviors. Inappropriate play decreased slightly and stereotypy decreased to very low levels. Social interaction increased substantially in Experiment 2 when social play was targeted but little change was observed in Experiment 1 when only independent play was targeted. Treatment gains were maintained for 26 weeks in Experiment 1 and 10 weeks in Experiment 2. In addition, the treatment gains were generalized across subjects and settings in Experiment 2. Finally, regular follow-up checks showed that independent and social play remained in the repertoire of the subjects for 12 months following the termination of programmed maintenance.