Background People with severe mental illness die 15–20 years earlier than the general population. Reasons why include that people with severe mental illness are more likely to smoke and be physically inactive as a result of social inequalities. Objectives (1) Evaluate the clinical effectiveness of multiple risk behaviour interventions on behaviour change (e.g. smoking abstinence), and outcomes affected by behaviours (e.g. weight loss). (2) Compare the clinical effectiveness of interventions targeting multiple and single risk behaviours. (3) Examine the factors affecting outcomes (e.g. intervention content). (4) Assess the factors affecting experiences of interventions (e.g. barriers and facilitators). Data sources The Cochrane Central Register of Controlled Trials (CENTRAL), EMBASE™ (Elsevier, Amsterdam, the Netherlands), MEDLINE, PsycInfo^® (American Psychological Association, Washington, DC, USA) and Science Citation Index (Clarivate Analytics, Philadelphia, PA, USA) were searched from inception to October 2018, and an updated search was conducted in March 2020. An Applied Social Sciences Index and Abstracts (ASSIA) search and an updated Cochrane Central Register of Controlled Trials search were undertaken in September 2020. Study selection Randomised controlled trials targeting single or multiple health risk behaviours among people with severe mental illness were included. Qualitative evidence on factors affecting the effectiveness of risk behaviour interventions was included. Study appraisal Network meta-analyses were conducted to compare the effectiveness of multiple and single risk behaviour interventions. The mean differences were estimated for continuous outcomes; if this was not possible, standardised mean differences were calculated. Thematic syntheses of qualitative studies were conducted. Results A total of 101 studies (67 randomised controlled trials and 34 qualitative studies) were included. Most outcomes were smoking abstinence, weight and body mass index. Just over half of studies were rated as having a high overall risk of bias. Trials focusing on smoking alone led to greater abstinence than targeting smoking and other behaviours. However, heterogeneity means that other explanations cannot be ruled out. For weight loss and body mass index, single risk behaviour (e.g. physical activity alone) and multiple risk behaviour (e.g. diet and physical activity) interventions had positive but modest benefits. For example, any risk behaviour intervention led to a 2 kg greater weight loss (–2.10 kg, 95% credible interval –3.14 to –1.06 kg) and approximately half a point (i.e. 0.5 kg/m²) greater body mass index reduction (–0.49 kg/m², 95% credible interval –0.97 to –0.01 kg/m²) than treatment as usual. There were potential synergies for targeting multiple health behaviours for reduction in systolic and diastolic blood pressure. No evidence was found of a deterioration in mental health for people with severe mental illness engaging in interventions to reduce health risk behaviours. Qualitative studies found that people with severe mental illness favoured interventions promoting physical and mental health together, and that took their condition into account. However, trials focused mainly on promoting physical health. Limitations Most quantitative studies focused on weight and body mass index; few assessed behavioural outcomes. Qualitative studies often addressed different aims. Conclusions Multiple and single risk behaviour interventions were associated with positive but modest benefits on most outcomes. Interventions seeking to promote physical health were not associated with deterioration in mental health. There was a lack of overlap between quantitative and qualitative studies. Future work Further research is needed to investigate whether or not health behaviour changes are maintained long term; tailoring weight-loss interventions for people with severe mental illness; and in terms of methods, co-production and mixed-methods approaches in future trials. Study registration This study is registered as PROSPERO CRD42018104724. Funding This project was funded by the National Institute for Health Research (NIHR) Health and Social Care Delivery Research programme and will be published in full in Health and Social Care Delivery Research; Vol. 10, No. 6. See the NIHR Journals Library website for further project information.