The purpose of this study was to determine if an intervention consisting of contingently imitating play, modeling expansions of play actions, and describing play actions increased the diversity of object play in young children with disabilities. The multicomponent intervention was introduced in a multiple-probe design across three young children in their classrooms. Generalization probe sessions, using an untrained toy set, were conducted throughout baseline and intervention. Follow-up sessions were conducted approximately 1 month after completion of the intervention. All participants increased their performance of different actions and the complexity of their play with toys; however, the magnitude of effects varied across participants. In addition, all participants spoke more and used more different words at the end of intervention when compared to the end of baseline. Performance during generalization and maintenance assessments was variable. Implications for future research and practice are discussed. (Contains 3 tables and 2 figures.)