Stepped, evidence-based and integrated care service models have the potential to be used as a reference for mental health services. RECOVER aimed to evaluate cost savings, effectiveness, and cost-effectiveness of such a model within a two arm, assessor- and data analysist-blinded RCT in Hamburg, Germany. Participants aged 16–79 years with mental disorders were randomly assigned either to RECOVER or treatment as usual (TAU). Primary outcomes comprised costs, effectiveness (combined symptoms, functioning, quality of life), and cost-effectiveness, hierarchically ordered. Outcomes were evaluated according to the ITT principle, group differences regarding costs with adjusted generalized linear models, effectiveness with ANCOVA models, and cost-effectiveness with the incremental cost-effectiveness ratio (ICER) and cost-effectiveness acceptability curves (CEACs). Between 1/1/2018 and 12/31/2020, n = 891 were finally included (n = 477 in RECOVER, n = 444 in TAU). RECOVER was associated with significantly lower annual total costs (-22 %), health and social care costs (-25 %) and hospital costs (-50 %). Effectiveness analyses showed a significantly better outcome for RECOVER with the fully imputed data . The CEACs descriptively demonstrated that RECOVER was cost-effective with a probability of >95 %. Treatment in RECOVER resulted in substantial cost reductions with better cost-effectiveness. RECOVER can be recommended as a reference model for comprehensive and integrated mental health services.