Background Obesity is twice as common in people with schizophrenia as in the general population. The National Institute for Health and Care Excellence guidance recommends that people with psychosis or schizophrenia, especially those taking antipsychotics, be offered a healthy eating and physical activity programme by their mental health care provider. There is insufficient evidence to inform how these lifestyle services should be commissioned. Objectives To develop a lifestyle intervention for people with first episode psychosis or schizophrenia and to evaluate its clinical effectiveness, cost-effectiveness, delivery and acceptability. Design A two-arm, analyst-blind, parallel-group, randomised controlled trial, with a 1 : 1 allocation ratio, using web-based randomisation; a mixed-methods process evaluation, including qualitative case study methods and logic modelling; and a cost–utility analysis. Setting Ten community mental health trusts in England. Participants People with first episode psychosis, schizophrenia or schizoaffective disorder. Interventions Intervention group: (1) four 2.5-hour group-based structured lifestyle self-management education sessions, 1 week apart; (2) multimodal fortnightly support contacts; (3) three 2.5-hour group booster sessions at 3-monthly intervals, post core sessions. Control group: usual care assessed through a longitudinal survey. All participants received standard written lifestyle information. Main outcome measures The primary outcome was change in weight (kg) at 12 months post randomisation. The key secondary outcomes measured at 3 and 12 months included self-reported nutrition (measured with the Dietary Instrument for Nutrition Education questionnaire), objectively measured physical activity measured by accelerometry [GENEActiv (Activinsights, Kimbolton, UK)], biomedical measures, adverse events, patient-reported outcome measures and a health economic assessment. Results The trial recruited 414 participants (intervention arm: 208 participants; usual care: 206 participants) between 10 March 2015 and 31 March 2016. A total of 341 participants (81.6%) completed the trial. A total of 412 participants were analysed. After 12 months, weight change did not differ between the groups (mean difference 0.0 kg, 95% confidence interval –1.59 to 1.67 kg; p = 0.964); physical activity, dietary intake and biochemical measures were unchanged. Glycated haemoglobin, fasting glucose and lipid profile were unchanged by the intervention. Quality of life, psychiatric symptoms and illness perception did not change during the trial. There were three deaths, but none was related to the intervention. Most adverse events were expected and related to the psychiatric illness. The process evaluation showed that the intervention was acceptable, with participants valuing the opportunity to interact with others facing similar challenges. Session feedback indicated that 87.2% of participants agreed that the sessions had met their needs. Some indicated the desire for more ongoing support. Professionals felt that the intervention was under-resourced and questioned the long-term sustainability within current NHS settings. Professionals would have preferred greater access to participants’ behaviour data to tailor the intervention better. The incremental cost-effectiveness ratio from the health-care perspective is £246,921 per quality-adjusted life-year (QALY) gained and the incremental cost-effectiveness ratio from the societal perspective is £367,543 per QALY gained. Conclusions Despite the challenges of undertaking clinical research in this population, the trial successfully recruited and retained participants, indicating a high level of interest in weight management interventions; however, the STEPWISE intervention was neither clinically effective nor cost-effective. Further research will be required to define how overweight and obesity in people with schizophrenia should be managed. The trial results suggest that lifestyle programmes for people with schizophrenia may need greater resourcing than for other populations, and interventions that have been shown to be effective in other populations, such as people with diabetes mellitus, are not necessarily effective in people with schizophrenia. Trial registration Current Controlled Trials ISRCTN19447796. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 22, No. 65. See the NIHR Journals Library website for further project information.