We evaluated the predictive value of resting pulmonary function tests (PFTs) in the determination of maximal exercise capacity in patients with obstructive and restrictive ventilatory disease. We performed resting PFTs and an incremental exercise study on a bicycle ergometer in 146 patients with chronic obstructive pulmonary disease (COPD) and 68 patients with restrictive disease. The patients with obstructive disease were further subdivided into mild, moderate, and severe based on the severity of their airway obstruction (mean +/- SD:FEV1, 2.78 +/- 0.77, 2.12 +/- 0.74, and 1.06 +/- 0.47, respectively). Correlation coefficients for PFTs vs VO2max and VEmax in restricted patients was generally low (DL = 0.67 and 0.34, IC = 0.58 and 0.35, FVC = 0.57 and 0.35, TLC = 0.35 and 0.18). In patients with COPD, the maximum voluntary ventilation (MVV) was the single best predictor of VO2 in all groups with correlation coefficients of 0.75, 0.69, and 0.89 in the mild, moderate, and severe subgroups, respectively. Similarly, the MVV was the best predictor of VEmax in all groups with correlation coefficients of 0.59, 0.64, and 0.89 in the three subgroups. The correlation with FEV1 was slightly less for both VO2max (0.69, 0.65, and 0.87) and VEmax (0.52, 0.64, 0.64) in the mild, moderate, and severe subgroups, respectively. Our findings show that PFTs are unreliable in predicting VEmax and VO2max in restricted patients. In patients with obstruction, the MVV is the single best predictor of VO2max and VEmax in all three categories, but was not significantly improved by stepwise multiple regression with additional PFT variables. Higher correlations were obtained in the severe group in whom the correlation with VO2max and VEmax was 0.89. However, the 95 percent confidence interval of the estimate for VO2 and VE was relatively large (+/- 0.16 L/min and +/- 6.6 L/min, respectively). We conclude that although several PFTs correlate significantly with maximum exercise, the large variance precludes their use to accurately predict maximum performance in individual patients with COPD.