A theoretical model for oxygen transport assuming a series linkage of ventilation, diffusion, oxygen uptake by erythrocytes, cardiac output, and oxygen release was used to calculate expected values for maximal oxygen intake (VO2max) of patients with various pulmonary disorders 22 patients with either restrictive or obstructive ventilatory impairment were studied at rest and maximal exercise. When exercise measurements of maximal pulmonary blood flow (QCmax), oxygen capacity, membrane diffusing capacity for CO, pulmonary capillary blood volume, alveolar ventilation, and mixed venous oxygen saturation were employed as input values, predictions of VO2max from the model correlated closely with measured values (r = 0.978). Measured VO2max was 976+/-389 ml/min (45.3+/-13% of predicted normal), and VO2max predicted from the model was 1,111+/-427 ml/min. The discrepancy may in part reflect uneven matching of alveolar ventilation, pulmonary capillary blood flow, and membrane diffusing capacity for CO within the lung; uniform matching is assumed in the model so that mismatching will impair gas exchange beyond our predictions. Although QCmax was less than predicted in most patients (63.6+/-19.6% of predicted) the model suggests that raising QCmax to normal could have raised VO2max only 11.6+/-8.8% in the face of existent impairment of intrapulmonary gas exchange. Since pulmonary functions measured at rest correlated well with exercise parameters needed in the model to predict VO2max we developed a nomogram for predicting VO2max from resting CO diffusing capacity, the forced one second expired volume, and the resting ratio of dead space to tidal volume. The correlation coefficient between measured and predicted VO2max, by using this nomogram, was 0.942.