To evaluate whether maximal oxygen consumption (VO2 max) measured by cardiopulmonary exercise test (CAR-PET) reflects cardiac reserve in patients with congestive heart failure (CHF), supine bicycle CAR-PET and exercise hemodynamic measurements were performed simultaneously in 12 patients with CHF of NYHA II-IV. With increasing workload, VO2 and cardiac output elevated gradually, then plateaued, demonstrating that patients with CHF could reach VO2max. According to VO2max, patients were divided into 4 classes: including 2 patients of class A (VO2max: 24.5 +/- 2.29 ml.min-1/kg, mean +/- s mean), 3 of B (17.6 +/- 1.37 ml.min-1/kg), 5 of C (13.6 +/- 0.66 ml.min-1/kg) and 2 of D (6.5 +/- 1.64 ml.min-1/kg). Maximal cardiac indices were 8.79 +/- 2.35 L.min-1/m2 in class A, 5.82 +/- 0.97 L.min-1/m2 in B, 3.53 +/- 0.95 L.min-1/m2 in C and 2.21 +/- 1.56 L.min-1/m2 in D. No significant correlation between supine resting hemodynamic parameters and VO2max/kg was found, suggesting that exercise tolerance could not be predicted by the measurement of resting cardiac performance. Furthermore, VO2max correlated poorly with NYHA classification in these patients. However, cardiac output correlated linearly with VO2 during exercise, suggesting that VO2 max/kg is a good predictor for cardiac reserve in CHF(CI = 0.6809 +/- 0.2748 VO2/kg, n = 40, r = 0.84, P < 0.0001; CO = 1.1618 +/- 7.9065 VO2, n = 40, r = 0.84, P < 0.0001). The results also showed that VO2max/kg did not correlate with the changes of pulmonary capillary wedge pressure (PCWP), indicating that exercise tolerance in CHF depends more on cardiac output than on ventilatory consequence of pulmonary congestion.(ABSTRACT TRUNCATED AT 250 WORDS)