To investigate the mechanisms of pulmonary gas-exchange impairment in idiopathic pulmonary fibrosis (IPF) and to evaluate their potential relationship to the CO diffusing capacity (DlCO)> we studied 15 patients with IPF (mean DlCO, 52% of predicted) at rest (breathing room air and pure O2) and during exercise. We measured pulmonary hemodynamics and respiratory gas-exchange variables, and we separated the ventilation-perfusion (a) mismatching and O2 diffusion limitation components of arterial hypoxemia using the multiple inert gas elimination technique. At rest a/ mismatching was moderate (2 to 4% of cardiac output perfusing poorly or unventilated lung units), and 19% of AaPO2 was due to O2 diffusion limitation. During exercise a/ mismatch did not worsen but the diffusion component of arterial hypoxemia increased markedly (40% AaPO2, p < 0.005). We observed that those patients with higher pulmonary vascular tone (more release of hypoxic pulmonary vasoconstriction) showed less pulmonary hypertension during exercise (p < 0.05), less a/ mismatching [at rest (p < 0.005) and during exercise (p < 0.0025)], and higher arterial PO2 during exercise (p = 0.01). We also found that DlCO corrected for alveolar volume (Kco) correlated with the mechanisms of hypoxemia during exercise [a/ mismatching (p < 0.025) and O2 diffusion limitation (p < 0.05)] and with the increase in pulmonary vascular resistance elicited by exercise (p < 0.005). In conclusion, we showed that the abnormalities of the pulmonary vasculature are key to modulate gas exchange in IPF, especially during exercise. We recommend the routine correction of DlCO for alveolar volume (Kco) in the clinical assessment of patients with IPF because the latter seems to be a useful functional indicator of both the severity of gas-exchange impairment during exercise and the degree of pulmonary vascular involvement.