Background: Self-harm in adolescents is common and repetition rates high. There is limited evidence of the effectiveness of interventions to reduce self-harm.Objectives: To assess the clinical effectiveness and cost-effectiveness of family therapy (FT) compared with treatment as usual (TAU). Design: A pragmatic, multicentre, individually randomised controlled trial of FT compared with TAU. Participants and therapists were aware of treatment allocation; researchers were blind to allocation. Setting: Child and Adolescent Mental Health Services (CAMHS) across three English regions. Participants: Young people aged 11–17 years who had self-harmed at least twice presenting to CAMHS following self-harm. Interventions: Eight hundred and thirty-two participants were randomised to manualised FT delivered by trained and supervised family therapists (n = 415) or to usual care offered by local CAMHS following self-harm (n = 417). Main outcome measures: Rates of repetition of self-harm leading to hospital attendance 18 months after randomisation. Results: Out of 832 young people, 212 (26.6%) experienced a primary outcome event: 118 out of 415 (28.4%) randomised to FT and 103 out of 417 (24.7%) randomised to TAU. There was no evidence of a statistically significant difference in repetition rates between groups (the hazard ratio for FT compared with TAU was 1.14, 95% confidence interval 0.87 to 1.49; p = 0.3349). FT was not found to be cost-effective when compared with TAU in the base case and most sensitivity analyses. FT was dominated (less effective and more expensive) in the complete case. However, when young people’s and caregivers’ quality-adjusted life-year gains were combined, FT incurred higher costs and resulted in better health outcomes than TAU within the National Institute for Health and Care Excellence cost-effectiveness range. Significant interactions with treatment, indicating moderation, were detected for the unemotional subscale on the young person-reported Inventory of Callous–Unemotional Traits (p = 0.0104) and the affective involvement subscale on the caregiver-reported McMaster Family Assessment Device (p = 0.0338). Caregivers and young people in the FT arm reported a range of significantly better outcomes on the Strengths and Difficulties Questionnaire. Self-reported suicidal ideation was significantly lower in the FT arm at 12 months but the same in both groups at 18 months. No significant unexpected adverse events or side effects were reported, with similar rates of expected adverse events across trial arms. Conclusions: For adolescents referred to CAMHS after self-harm, who have self-harmed at least once before, FT confers no benefits over TAU in reducing self-harm repetition rates. There is some evidence to support the effectiveness of FT in reducing self-harm when caregivers reported poor family functioning. When the young person themselves reported difficulty expressing emotion, FT did not seem as effective as TAU. There was no evidence that FT is cost-effective when only the health benefits to participants were considered but there was a suggestion that FT may be cost-effective if health benefits to caregivers are taken into account. FT had a significant, positive impact on general emotional and behavioural problems at 12 and 18 months. Limitations: There was significant loss to follow-up for secondary outcomes and health economic analyses; the primary outcome misses those who do not attend hospital following self-harm; and the numbers receiving formal FT in the TAU arm were higher than expected.