Mechanisms of hypoxemia and hypocapnia in pulmonary embolism (PE) are incompletely understood. We studied 10 patients at diagnosis (D) and five of these again after 10 to 14 d of heparin treatment (T). Patients had right heart catheterization, assessment of ventilation-perfusion ratio (VA/Q) distribution by inert gas, radioisotopic perfusion and ventilation scans, and angiography. At D, two-thirds of the pulmonary circulation was obstructed, patients were hypoxemic (PaO2 = 63.0 +/- 11.7 mm Hg) and hypocapnic (PaCO2 = 30.0 +/- 4.1 mm Hg), mixed venous oxygen pressure (PvO2) was reduced (30.9 +/- 3.9 mm Hg), minute ventilation (VE) markedly increased (14.1 +/- 5.1 L/min), and cardiac output measured by applying the Fick principle to arteriovenous oxygen content difference (QT) slightly low (4.7 +/- 1.7 L/min). Hypoxemia was mainly explained by VA/Q inequality, reduced PvO2 also contributed. Hypocapnia was the result of hyperventilation. VA/Q inequality was characterized by shift of VA and Q distribution mean to regions with higher VA/Q ratio through a fraction of blood flow (19.0 +/- 24.3% of cardiac output) went to lung units with low VA/Q ratio. Log SDQ and log SDvA were increased. Shunt, diffusion limitation, or true alveolar dead space occurred in occasional patients but were generally insignificant. Regional ventilation and perfusion maps indicated that in the unperfused lung segments, ventilation was reduced. Furthermore, they disclosed overperfused lung segments. At T, hypoxemia and hypocapnia improved considerably. However, temporal imbalances in recovery between regional ventilation and perfusion occurred with the former normalizing sooner. However, perfusion recovered sooner than ventilation in some regions.