Aims We aimed to determine the differences of impact of cardiopulmonary exercise testing (CPX) parameters on prognosis of heart failure with reduced left ventricular ejection fraction (HFrEF), preserved ejection fraction (HFpEF) and mid-range ejection fraction (HFmrEF). Methods We compared clinical characteristics and CPX parameters among the three groups, and the value of each CPX parameter to predict adverse cardiac events (cardiac deaths and re-hospitalizations for heart failure), cardiac deaths and all-cause deaths. Results Of 1190 patients, 41.9% had HFrEF, 36.8% had HFpEF and 21.3% had HFmrEF. The patients in HFrEF group had higher rates of adverse cardiac events, cardiac death and all-cause death than those of HFpEF and HFmrEF groups. In HFrEF, the independent predictors of adverse cardiac events were peak oxygen consumption and oxygen uptake efficiency slope, predictors of cardiac death were peak oxygen consumption and oxygen uptake efficiency slope, and the predictor of all-cause death was peak oxygen consumption. In HFpEF, the predictor of adverse cardiac events was peak oxygen consumption, predictors of cardiac deaths and all-cause deaths were peak oxygen consumption and exertional oscillatory ventilation. In HFmrEF, predictors of adverse cardiac events were peak oxygen consumption and oxygen uptake efficiency slope, and the predictor of cardiac deaths and all-cause deaths was peak oxygen consumption. Conclusion Peak oxygen consumption is the strong predictor for adverse events in all groups. Oxygen uptake efficiency slope predicts adverse prognosis in HFrEF, but not in HFpEF. In contrast, exertional oscillatory ventilation is the predictor only in HFpEF. Thus, different CPX parameters may be able to differentially predict prognosis in HFrEF and HFpEF. Those for predicting prognosis in HFmrEF may be intermediate between HFrEF and HFpEF.