This review critically explores the development, impact and evaluation of exercise referral schemes (ERS) in the UK. A rapid expansion in the use of such ERSs has been recorded throughout leisure and primary care settings, but the evidence underpinning their implementation has been sparse and predominantly limited to randomized control trial (RCT) research design. Consequently, understanding of exercise referral as a 'real world' intervention has been limited. Considering the increasing importance being placed on evidence-based practice and clinical effectiveness, it is no longer sufficient for service providers of exercise referral to ignore the need to evaluate schemes. The guidelines on evaluation provided by the National Quality Assurance Framework for Exercise Referral are limited, hence practitioners are often unsure of the best measures to use when assessing effectiveness. Predominantly, exercise professionals focus on the collection of physiological data but tend to ignore relevant psychological and environmental parameters. Also, few UK studies have followed participants up in the long term, to see if physical activity behaviour is sustained over time. Here, evidence from two on-going, large-scale (n = 1600/annum) evaluation studies of exercise referral schemes, based in urban localities in the northwest of England, are described. A participatory action research framework for evaluation was utilized and incorporated multi-method research approaches for the assessment of both ERS participants and health professionals involved in intervention delivery. This framework is an appropriate methodology for the evaluation and development of complex interventions, and here incorporates case study, focus groups, interviews and survey questionnaires. Included was a 12-month tracking study of a cohort of exercise referral participants (n = 342), which measured leisure-time physical activity levels (Godin leisure time score), at baseline (entry to exercise referral) and at 3 monthly intervals thereafter. Adherence to the ERS was approximately 35-45%, with the older participants more likely to complete. Physiological changes during the ERS, although statistically significant, were not of a magnitude to convey any real health benefit to an individual's health status. Although small in scale, physiological changes were all in a positive direction (e.g. reduction in blood pressure) and, if maintained over time, could bring about population-level benefits in health. Participants referred from cardiac and practice nurses had higher levels of adherence than participants referred by general practitioners. Scheme B showed that the participants who adhered (n = 103) until the end of the ERS (12 weeks) were able to sustain a small increase in physical activity at the end of 12 months (increase of 21 min moderate activity/week compared with baseline). In conclusion, this research shows that the process of exercise referral benefits certain segments of the population, but not necessarily all.