Abstract
Two elementary school children who showed low levels of social interaction and play activities with other children were taught to say three statements relevant to pro‐social behaviour. The children were then praised for achieving correspondence between what they said they would do and actual performance of the target behaviour during an unstructured classroom activities session. Treatment was evaluated using a repeated measures multiple baseline across subjects design. The two subjects initially interacted with peers approximately 8 percent of the time during baseline and subsequently 50 percent of the time during the treatment phase, equating them with the class mean for interactions occurring across 50 percent of the available time. Follow‐up observations after nine weeks showed treatment effects to have been maintained.