Background People with psychotic disorders have reduced life expectancy largely because of physical health problems, especially cardiovascular disease, that are complicated by the use of tobacco and cannabis. Objectives We set out to (1) chart lifestyle and substance use choices and the emergence of cardiometabolic risk from the earliest presentation with psychosis, (2) develop a pragmatic health promotion intervention integrated within the clinical teams to improve the lifestyle choices and health outcomes of people with psychosis and (3) evaluate the clinical effectiveness and cost-effectiveness of that health promotion intervention. Design We performed a longitudinal cohort study of people presenting with their first episode of psychosis in three mental health trusts and followed up participants for 1 year [work package 1, physical health and substance use measures in first episode of psychosis (PUMP)]. We used an iterative Delphi methodology to develop and refine a modular health promotion intervention, improving physical health and reducing substance use in psychosis (IMPaCT) therapy, which was to be delivered by the patient’s usual care co-ordinator and used motivational interviewing techniques and cognitive–behavioural therapy to improve health choices of people with psychosis (work package 2). We then conducted a multicentre, two-arm, parallel-cluster, randomised controlled trial to determine the clinical effectiveness and cost-effectiveness of using the intervention with people with established psychosis (work package 3: IMPaCT randomised controlled trial) in five UK mental health trusts. The work took place between 2008 and 2014. Participants All people aged between 16 and 65 years within 6 months of their first presentation with a non-organic psychosis and who were proficient in English were eligible for inclusion in the PUMP study. Participants in the work package 2 training development were staff selected from a range of settings, working with psychosis. Participants in the phase 3 Delphi consensus and manual development comprised three expert groups of (1) therapists/researchers recruited from the local and national community, (2) clinicians and (3) service users, each of whom took part in two iterative review and feedback sessions. For work package 3, IMPaCT randomised controlled trial, care co-ordinators in participating community mental health teams who were permanently employed and had a minimum of four eligible patients (i.e. aged between 18 and 65 years with a diagnosis of a psychotic disorder) on their caseload were eligible to participate. In studies 1 and 3, patient participants were ineligible if they were pregnant or had a major illness that would have had an impact on their metabolic status or if they had a significant learning disability. All participants were included in the study only after giving written confirmed consent. Main outcome measures Cardiometabolic risk markers, including rates of obesity and central obesity, and levels of glycated haemoglobin (HbA_1c) and lipids, were the main outcomes in work package 1 (PUMP), with descriptive data presented on substance use. Our primary outcome measure for the IMPaCT randomised controlled trial was the physical or mental health component Short Form questionnaire-36 items quality-of-life scores at 12 months. Results Obesity rates rose from 18% at first presentation with psychosis to 24% by 1 year, but cardiometabolic risk was not associated with baseline lifestyle and substance use choices. Patterns of increase in the levels of HbA_1c over the year following first presentation showed variation by ethnic group. We recruited 104 care co-ordinators, of whom 52 (with 213 patients) were randomised to deliver IMPaCT therapy and 52 (with 193 patients) were randomised to deliver treatment as usual, in keeping with our power calculations. Of these 406 participants with established psychosis, 318 (78%) and 301 (74%) participants, respectively, attended the 12- and 15-month follow-ups. We found no significant effect of IMPaCT therapy compared with treatment as usual on the physical or mental health component Short Form questionnaire-36 items scores at either time point in an intention-to-treat analysis [physical health score (‘d’) –0.17 at 12 months and –0.09 at 15 months; mental health score (‘d’) 0.03 at 12 months and –0.05 at 15 months] or on costs. Nor did we find an effect on other cardiovascular risk indicators, including diabetes, except in the case of high-density lipoprotein cholesterol, which showed a trend for greater benefit with IMPaCT therapy than with treatment as usual (treatment effect 0.085, 95% confidence interval 0.007 to 0.16; p = 0.034). Limitations Follow-up in work package 1 was challenging, with 127 out of 293 participants attending; however, there was no difference in cardiometabolic measures or demographic factors at baseline between those who attended for follow-up and those who did not. In work package 3, the IMPaCT randomised controlled trial, care co-ordinators struggled to provide additional time to their patients that was devoted to the health promotion intervention on top of their usual clinical care contact with them. Conclusions Cardiometabolic risk is prominent even soon after first presentation with psychosis and increases over time. Lifestyle choices and substance use habits at first presentation do not predict those who will be most cardiometabolically compromised 1 year later. Training and supervising care co-ordinators to deliver a health promotion intervention to their own patients on top of routine care is not effective in the NHS for improving quality of life or reducing cardiometabolic risk. Future work Further work is needed to develop and evaluate effective, cost-effective and affordable ways of preventing the emergence of and reversing existing cardiometabolic risk indicators in people with psychosis. Trial registration Current Controlled Trials ISRCTN58667926. Funding This project was funded by the National Institute for Health Research (NIHR) Programme Grants for Applied Research programme and will be published in full in Programme Grants for Applied Research; Vol. 8, No. 1. See the NIHR Journals Library website for further project information.