The study describes a training program in which young children with severe and moderate handicaps were taught to generalize play responses to multiple sets of toys. A multiple probe design, replicated with four children, was used to assess the effects of generalization training within four sets of toys on generalization to untrained toys from four other sets. The responses taught were unique to each set of toys. Results indicated that training to generalize wihin two sets of toys was associated with stimulus generalization of other sets that did not formerly show generalization in three participants. Probes were also taken on responses to two additional sets of toys that differed from the previous sets in topography and in the effects that the toys produced. While the participants generalized to between 50% and 100% of the toys that were similar in responses and effects they did not generalize to toys from the dissimilar sets. Implications for conducting research using strategies based on response interrelationships in training contexts are discussed. (Author)