Several interventions have demonstrated efficacy in improving social outcomes for children with autism; however, few have been successfully implemented in schools. This study compared two implementation strategies to improve the use of a social engagement intervention for children with autism in public schools. In total, 31 children with autism in grades K-5 and 28 school personnel participated in a randomized controlled trial. Schools were randomized to (1) training in Remaking Recess, a social engagement intervention, or (2) training in Remaking Recess with implementation support. Linear regression with random effects was used to test the intervention effects on implementation fidelity and social outcomes (peer engagement, social network inclusion, and friendship nominations). In both groups, implementation fidelity improved after training but remained low. Children in the Remaking Recess with implementation support condition had significantly higher social network inclusion and received more friendship nominations than children in the Remaking Recess-only condition (p = 0.03). Children in both groups experienced reduced solitary engagement (p < 0.001) and increased joint engagement (p < 0.001). The results suggest that implementation supports may have an effect on outcomes above and beyond the intervention, and that further research is needed into the active intervention mechanisms.