Objectives To assess the cost-effectiveness of LEGO-based therapy compared with usual support. Design Cost-utility analysis alongside randomised control trial. Setting Mainstream primary and secondary schools in the UK. Participants 248 children and young people (CYP) with autism spectrum disorder (ASD) aged 7–15 years. Intervention LEGO-based therapy is a group social skills intervention designed specifically for CYP with ASD. Through play, CYP learn to use the skills such as joint attention, sharing, communication and group problem-solving. CYP randomised to the intervention arm received 12 weekly sessions of LEGO-based therapy and usual support, while CYP allocated to control arm received usual support only. Main outcome measures Average costs based on National Health Service (NHS) and personal social services perspective and quality-adjusted life years (QALYs) measured by EQ-5D-Y over time horizon of 1 year were collected during the trial. Incremental cost-effectiveness ratio (ICER) was calculated, and non-parametric bootstrapping was conducted. The uncertainty around the ICER estimates was presented using cost-effectiveness acceptability curve (CEAC). A set of sensitivity analyses were conducted to assess the robustness of the primary findings. Results After adjustment and bootstrapping, on average, CYP in LEGO-based therapy group incurred less costs (incremental cost was −£251 (95% CI −£752 to £268)) and gained marginal improvement in QALYs (QALYs gained 0.009 (95% CI −0.008 to 0.028)). The CEAC shows that the probability of LEGO-based therapy being cost-effective was 94% at the willingness-to-pay threshold of £20 000 per QALY gained. Results of sensitivity analyses were consistent with the primary outcomes. Conclusion Compared with usual support, LEGO-based therapy produced marginal reduction in costs and improvement in QALYs. Results from both primary and sensitivity analyses suggested that LEGO-based therapy was likely to be cost-effective. Trial registration number ISRCTN64852382.