Objectives:  The purpose of this updated meta-analysis was to determine whether exercise training reversed left ventricular remodelling in heart failure patients. Methods:  Articles were searched in MEDLINE, Cochrane Central Register of Controlled Trials, CINAHL, EMBASE and PubMed (up until October, 2011). The anti-remodelling benefit of exercise training was assessed by ejection fraction (EF), end-diastolic volume (EDV) and end-systolic volume (ESV). Standardised mean difference (SMD) together with 95% confidence intervals (CI) was calculated. Subgroup meta-analysis with exercise duration was also conducted. Results:  Fifteen randomised controlled trials with 813 patients were included. Aerobic exercise training improved EF (SMD = 0.44; 95% CI 0.28 to 0.61), EDV (SMD = -0.33; 95% CI -0.49 to -0.16) and ESV (SMD = -0.40; 95% CI -0.57 to -0.23). Subgroup analysis indicated that long-term aerobic exercise (≥6 months) had a marked positive effect on EF (SMD = 0.5; 95% confidence interval 0.31 to 0.69), EDV (SMD = -0.38; 95% CI -0.57 to -0.19) and ESV (SMD = -0.48; 95% CI -0.67 to -0.29), but there was no evidence of benefit with short-term aerobic exercise (<6 months): EF (SMD = 0.27; 95% CI -0.08 to 0.61), EDV (SMD = -0.14; 95% CI -0.48 to 0.21) and ESV (SMD = -0.08; 95% CI -0.47 to 0.30). Strength training (alone or plus aerobic training) was not associated with improvements in EDV and ESV, with all confidence intervals including 0. Conclusions:  Aerobic exercise training, especially long-term duration (≥6 months) reverses left ventricular remodelling in clinically stable patients with heart failure. Strength training (alone or plus aerobic training) did not improve or worsen ventricular remodelling.