Purpose: A prior randomized trial found a school social intervention yielded significantly better outcomes (social and autism features) immediately following intervention compared to typical school programming (services-as-usual [SAU]) for children on the autism spectrum. In that study, children in the SAU condition subsequently completed a summer social intervention. This study tested longer-term maintenance of effects for children who completed both interventions. Methods: A total of 103 children (ages 6-12 years) on the autism spectrum enrolled and 102 completed the initial RCT. Following the summer social intervention, 90 children from the original RCT completed the longer-term follow-up study. In addition to baseline and posttest in the initial RCT, children from both groups were tested at three follow-up points (five total testing points). At the time of first longitudinal follow-up testing, the children were 1.25-4.25 years post-intervention (ages 8-15 years). Results: Longitudinal multilevel model analyses (and follow-up contrasts) revealed significant improvements for both groups post-intervention on measures of emotion recognition, autism features, and social skills, indicating maintenance of post-intervention improvements over the three follow-up testing points. No between-group differences were found for autism features or social skills over time; however, the school social intervention may have yielded somewhat better emotion recognition skills. Exploratory tests found that child IQ, language level, and length of time since completing the intervention did not moderate outcomes. Conclusion: Both social interventions yielded positive and durable longer-term improvements for children on the autism spectrum. [This paper will be published in "Journal of Autism and Development Disorders."]