BackgroundCognitive behavioural therapy (CBT) is effective in treating various neurological and psychiatric diseases. It improves anxiety symptoms in children with autism spectrum disorder, gaining considerable empirical support. However, social skills results are mixed, leading to debate over its effectiveness, highlighting the need for further development. While the Social Responsiveness Scale (SRS) is a secondary indicator to measure anxiety symptoms, it primarily evaluates social skills, which are essential for rehabilitating children with autism. Therefore, evaluating social disorder improvement in children with autism is imperative. Social impairment is a core autism symptom. Therefore, we conducted a systematic review of randomised controlled trials assessing the effects of CBT on social skills in this population. MethodsWe reviewed articles published in several databases through October 2022 and relevant reference lists. We used the standardised mean difference (SMD) as the main effect size indicator and focused on SRS metrics from baseline to endpoint. We analysed subgroups, heterogeneity, bias risk, and publication bias. ResultsOur meta-analysis included 214 children from seven randomised controlled trials with nine datasets. Forest plot analysis shows CBT improved social skills in children with autism compared to controls. Subgroup analysis revealed parents' and teachers' SRS scores for children, SRS scores of CBT versus waitlist controls, and those of CBT versus non-waiting-list controls. LimitationsMost randomised controlled CBT trials for children with autism have explored anxiety symptom improvement. Further, social skill assessment was a secondary outcome or not assessed. Thus, social skills data are insufficient. ConclusionsCBT is effective in improving social impairment in children with autism. RegistrationThis meta-analysis was registered with the International Prospective Register of Systematic Reviews (CRD42022363423).