Group social skills interventions (GSSIs) are a commonly offered treatment for children with high functioning ASD. We critically evaluated GSSI randomised controlled trials for those aged 6-25 years. Our meta-analysis of outcomes emphasised internal validity, thus was restricted to trials that used the parent-report social responsiveness scale (SRS) or the social skills rating system (SSRS). Large positive effect sizes were found for the SRS total score, plus the social communication and restricted interests and repetitive behaviours subscales. The SSRS social skills subscale improved with moderate effect size. Moderator analysis of the SRS showed that GSSIs that include parent-groups, and are of greater duration or intensity, obtained larger effect sizes. We recommend future trials distinguish gains in children's social knowledge from social performance.