Despite the popularity of social skills groups, there remains a need for empirical investigation of treatment effects, especially when targeting pivotal aspects of social functioning such as initiations to peers. The goal of the present study was to conduct a randomized controlled trial of a 12-week social intervention (SUCCESS), which combined an inclusive social group with a parent education program. Twenty-five 4- to 6-year-olds with Autism Spectrum Disorder (ASD) were randomized to SUCCESS (N = 11) or to treatment as usual (N = 14). Combining a peer group model with a parent training program, the SUCCESS intervention used naturalistic behavioral techniques (e.g., environmental arrangement, natural reinforcement) to increase social initiations to peers. After 12 weeks, children participating in the SUCCESS program made more frequent initiations to peers than children in the treatment-as-usual group, including more prompted and unprompted initiations to request. Additional gains in clinician-rated social functioning were observed in children randomized to SUCCESS, while differential treatment effects were not detected in parent-rated measures. However, lower baseline social motivation was associated with greater parent-reported initiation improvement. This study provides preliminary support for the efficacy of a naturalistic, behavioral social skills intervention to improve peer initiations for children with ASD. The findings suggest that using a motivation-based social skills group was effective in increasing both prompted and spontaneous initiations to peers, and highlights the need for further research into the role of baseline social motivation in predicting social skills treatment response.