Abstract Background Exercise is a recognised element of health-care management of mental-health conditions. In primary health care, it has been delivered through exercise referral schemes (ERS). The National Institute for Health and Care Excellence has highlighted uncertainty regarding the effectiveness of ERS in improving exercise participation and health outcomes among those referred for mental-health reasons. This review aims, therefore, to evaluate ERSs for individuals who are referred specifically for mental-health reasons. Methods Studies were reviewed that assessed the effectiveness of ERSs in improving initiation of and/or adherence to exercise and/or their effectiveness in improving long-term participation in exercise and health outcomes among primary care patients who had been referred to the scheme for mental-health reasons. The data were extracted and their quality assessed. Data were analysed through a narrative synthesis approach. Results Nine studies met the eligibility criteria. Three assessed clinical effectiveness of the schemes, eight assessed ERS uptake and/or adherence to the exercise schedule, and two assessed the impact of the ERSs on long-term exercise levels. In one study, it was found that ERSs that were based in leisure centres significantly improved long-term symptoms in those who had been referred due to their mental ill health (P<0.05). ERSs that involved face-to-face consultations and telephone calls had the highest rates of mean uptake (91.5%) and adherence (71.7%), but a difference was observed between uptake/adherence in trials (86.8%/55.3%) and in routine practice (57.9%/37.2%). ERSs that included face-to-face consultations and telephone calls increased the amount of long-term physical activity that was undertaken by people who had been referred for mental-health reasons (P=0.003). Conclusions Uptake and effectiveness of ERSs for mental health conditions was related to programme content and setting with more effective programmes providing both face-to-face and telephone consultations. Good uptake of yoga among those referred for mental health reasons suggests that mindful exercise options should be investigated further. Existing ERSs could be improved through application of individual tailoring and the provision of more face-to-face consultations, and social support. Further research is required to identify the types of ERSs that are most clinically effective for those with mental ill health.