Most patients with severe, acute pulmonary embolism (PE) have arterial hypoxemia. To further define the respective roles of ventilation to perfusion (VA/Q) mismatch and intrapulmonary shunt in the mechanism of hypoxemia, we used both right heart catheterization and the six inert gas elimination technique in seven patients with severe, acute PE (mean vascular obstruction, 55 percent) and hypoxemia (mean PaO2, 67 +/- 11 mm Hg). None had previous cardiopulmonary disease, and all were studied within the first ten days of initial symptoms. Increased calculated venous admixture (mean QVA/QT 16.6 +/- 5.1 percent) was present in all patients. The relative contributions of VA/Q mismatching and shunt to this venous admixture varied, however, according to pulmonary radiographic abnormalities and the time elapsed from initial symptoms to the gas exchange study. Although all patients had some degree of VA/Q mismatch, the two patients studied early (ie, less than 48 hours following acute PE) had normal chest x-ray film findings and no significant shunt; VA/Q mismatching accounted for most of the hypoxemia. In the others a shunt (3 to 17 percent of cardiac output) was recorded along with radiographic evidence of atelectasis or infiltrates and accounted for most of the venous admixture in one. In all patients, a low mixed venous oxygen tension (27 +/- 5 mm Hg) additionally contributed to the hypoxemia. Our findings suggest that the initial hypoxemia of acute PE is caused by an altered distribution of ventilation to perfusion. Intrapulmonary shunting contributes significantly to hypoxemia only when atelectasis or another cause of lung volume loss develops.