Background There is a high prevalence of smoking among people with severe mental ill health (SMI). Helping people with SMI to quit smoking could improve their health and longevity, and reduce health inequalities. However, those with SMI are less likely to access and engage with routine smoking cessation services than the general population. Objectives To compare the clinical effectiveness and cost-effectiveness of a bespoke smoking cessation (BSC) intervention with usual stop smoking services for people with SMI. Design A pragmatic, two-arm, individually randomised controlled trial. Setting Primary care and secondary care mental health services in England. Participants Smokers aged ≥ 18 years with SMI who would like to cut down on or quit smoking. Interventions A BSC intervention delivered by mental health specialists trained to deliver evidence-supported smoking cessation interventions compared with usual care. Main outcome measures The primary outcome was self-reported, CO-verified smoking cessation at 12 months. Smoking-related secondary outcomes were self-reported smoking cessation, the number of cigarettes smoked per day, the Fagerström Test for Nicotine Dependence and the Motivation to Quit questionnaire. Other secondary outcomes were Patient Health Questionnaire-9 items, Generalised Anxiety Disorder Assessment-7 items and 12-Item Short-Form Health Survey, to assess mental health and body mass index measured at 6 and 12 months post randomisation. Results The trial randomised 526 people (265 to the intervention group, 261 to the usual-care group) aged 19 to 72 years (mean 46 years). About 60% of participants were male. Participants smoked between 3 and 100 cigarettes per day (mean 25 cigarettes per day) at baseline. The intervention group had a higher rate of exhaled CO-verified smoking cessation at 6 and 12 months than the usual-care group [adjusted odds ratio (OR) 12 months: 1.6, 95% confidence interval (CI) 0.9 to 2.8; adjusted OR 6 months: 2.4, 95% CI 1.2 to 4.7]. This was not statistically significant at 12 months (p = 0.12) but was statistically significant at 6 months (p = 0.01). In total, 111 serious adverse events were reported (69 in the BSC group and 42 in the usual-care group); the majority were unplanned hospitalisations due to a deterioration in mental health (n = 98). The intervention is likely (57%) to be less costly but more effective than usual care; however, this result was not necessarily associated with participants’ smoking status. Limitations Follow-up was not blind to treatment allocation. However, the primary outcome included a biochemically verified end point, less susceptible to observer biases. Some participants experienced difficulties in accessing nicotine replacement therapy because of changes in service provision. Efforts were made to help participants access nicotine replacement therapy, but this may have affected participants’ quit attempt. Conclusions People with SMI who received the intervention were more likely to have stopped smoking at 6 months. Although more people who received the intervention had stopped smoking at 12 months, this was not statistically significant. Future work Further research is needed to establish how quitting can be sustained among people with SMI. Trial registration Current Controlled Trials ISRCTN72955454. 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. 23, No. 50. See the NIHR Journals Library website for further project information.