We investigated the effects of an intervention using high-probability request sequences with embedded peer modeling to increase social interactions of children with autism in a classroom. The effects of the intervention on compliant responding to social requests and social behaviors were monitored using a single-subject multiple baseline design across children. Additionally, social validity regarding the intervention goals, procedures, and outcomes was measured by relevant consumers. The results of this study indicate that all three children's compliant responding to low-probability requests and social behaviors increased with the intervention and were maintained. Furthermore, the target children's social behaviors generalized to untrained peers and nontraining settings. The social validity results indicated a high level of consumer acceptability and usability among relevant consumers.