Objective The aim of this review was to explore attendance of UK exercise referral schemes (ERS), who attends them, why participants drop out of schemes and to compare evaluations of existing ERS with randomised controlled trials (RCTs). Design Systematic review. Method A search of major databases was conducted to identify studies investigating ERS interventions that were based in primary care in the UK, reported attendance-related outcomes and were published in peer- reviewed journals. Results Five evaluations of existing ERS and four RCTs met the inclusion criteria. Method of participant recruitment was the only marked difference between the two types of study. In RCTs and evaluations, rates of referral uptake and attendance were varied but comparable. Attendance was generally poor; approximately eighty per cent of participants who took up referral dropped out before the end of programmes. More women than men took up referral (60 vs. 40 per cent) but there was no evidence of higher attendance in women. None of the participant characteristics reported were consistently associated with attendance. Most of the reasons for attrition and negative comments from participants related to practical problems associated with attending leisure facilities. Conclusion The present review highlighted a high level of attrition in ERS. However, poor measurement and reporting of attendance, and inadequate participant profiling, prevented us from identifying which sections of the population were most likely to attend or drop out. Adequate data collection regimens, beginning at the point of referral would enable us to learn whom exactly ERS are proving successful for.