Social skills group training (SSGT) is widely used for intellectually able children and adolescents with autism spectrum disorder (ASD). Previous studies indicate small to moderate effects on social communication capacities. The duration of most available programs is relatively short, and extended training might lead to further improvement. This randomized controlled trial compared an extended 24-week version of the SSGT program KONTAKT with standard care. The weekly sessions gradually shifted in content from acquisition of new skills to real-world application of the acquired skills. A total of 50 participants with ASD (15 females; 35 males) aged 8–17 years were included. The study was conducted at two child and adolescent psychiatry outpatient units in Sweden. The primary outcome was the Social Responsiveness Scale–Second Edition (SRS-2) rated by parents and blinded teachers. Secondary outcomes included parent- and teacher-rated adaptive behaviors, trainer-rated global functioning and clinical severity, and self-reported child and caregiver stress. Assessments were made at baseline, posttreatment, and at 3-months follow-up. Parent-rated SRS-2 scores indicated large effects posttreatment [− 19.2; 95% CI − 29.9 to − 8.5; p < .001, effect size (ES) = 0.76], which were maintained at follow-up (− 20.7; 95% CI − 31.7 to − 9.7; p < .0001, ES = 0.82). These estimates indicate substantially larger improvement than previously reported for shorter SSGT.