The frequency of social initiations and satisfaction with interactions in three dyads, each consisting of 1 student with disabilities and 1 nondisabled peer, were assessed under two alternating conditions: Condition 1 assessed the interactions around a set of four trained computer games, and Condition 2 assessed interactions when students were playing a set of four untrained computer games. Training was conducted with a multiple baseline design across participants and was followed by social interaction probes using an alternating treatments design. The results indicated greater frequencies of social initiation by 5 of 6 participants, higher degrees of game satisfaction by all participants, and equal or higher degrees of peer satisfaction by 5 of 6 participants when playing trained games in comparison to untrained games.