Left ventricular (LV) diastolic dysfunction is associated with the pathophysiology of heart failure with preserved ejection fraction (HFpEF) and contributes importantly to exercise intolerance that results in a reduced quality of life (QOL) in HFpEF patients. Although the effects of exercise training on LV diastolic function, exercise capacity, or QOL in HFpEF patients have been examined in randomized clinical trials (RCTs), results are inconsistent due partly to limited power with small sample sizes. We aimed to conduct a meta-analysis of RCTs examining the effects of exercise training on LV diastolic function and exercise capacity as well as QOL in HFpEF patients. The search of electronic databases identified 8 RCTs with 436 patients. The duration of exercise training ranged from 12 to 24 weeks. In the pooled analysis, exercise training improved peak exercise oxygen uptake (weighted mean difference [95% CI], 1.660 [0.973, 2.348] ml/min/kg), 6-min walk distance (33.883 [12.384 55.381] m), and Minnesota Living With Heart Failure Questionnaire total score (9.059 [3.083, 15.035] point) compared with control. In contrast, exercise training did not significantly change early diastolic mitral annular velocity (weighted mean difference [95% CI], 0.317 [− 0.952, 1.587] cm/s), the ratio of early diastolic mitral inflow to annular velocities (− 1.203 [− 4.065, 1.658]), or LV ejection fraction (0.850 [− 0.128, 1.828] %) compared with control. In conclusion, the present meta-analysis suggests that exercise training improves exercise capacity and QOL without significant change in LV systolic or diastolic function in HFpEF patients.