BACKGROUND/OBJECTIVES Improving patients' cardiorespiratory fitness is an important therapeutic outcome in cardiac rehabilitation. The ability of cardiac rehabilitation to reduce mortality and morbidity has been evidenced through several meta-analyses. Whether cardiac rehabilitation can increase cardiorespiratory fitness and which factors may influence such gains are less well quantified. METHODS We performed detailed literature searches of electronic databases and manually searched papers concerning changes in cardiorespiratory fitness in cardiac rehabilitation patients. We performed random-effects meta-analysis of mean improvements in cardiorespiratory fitness and subgroup analyses to determine potential sources of heterogeneity. RESULTS Data from 31 studies produced 48 groups (n=3827) with a mean improvement of 1.55 (95% CI 1.21-1.89) METs, (p<0.001); equivalent to standardised effect size of ES=0.97 (95% CI 0.80-1.13). As this value was highly heterogeneous (Q=852, p<0.001) we performed subgroup analyses on the effect size data. Gains in fitness were highest in patients receiving >36 exercise sessions in studies where fitness was assessed using the Naughton Protocol. Patient characteristics associated with the highest fitness gains were age (being young) and sex (being male training in a male-only exercise group). Changes in fitness were unrelated to programme type (comprehensive or exercise-only), duration or study design. There was no association with patient's baseline fitness levels. CONCLUSION This is the first meta-analysis of changes in cardiovascular fitness in cardiac rehabilitation patients and shows clinically significant improvements in a large sample of patients from a variety of rehabilitation programmes. This analysis helps describe the characteristics of cardiac rehabilitation programmes which can increase patients' cardiorespiratory fitness.