The relationship between the magnitude of hypoxic pulmonary vasoconstriction (h.p.v.) and arterial oxygen tension (Pa,O2) has been studied in intact anaesthetized dogs. Radioactive 133Xe was infused continuously into the inferior vena cava. A tracheal divider made it possible to vary the inspired gas composition to each lung independently. With constant ventilation the 133Xe in the mixed expired gas from each lung was proportional to the blood flow to that lung. Unilateral ventilation of the left lung with 7% oxygen produced a diversion of blood flow away from the lung and a reduction in Pa,O2. Repeated hypoxic stimuli produced a progressively greater reduction in the blood flow to the hypoxic lung and a progressive increase in Pa,O2. Administration of a beta-adrenergic agonist, dobutamine hydrochloride, during ventilation of the left lung with 7% oxygen, resulted in an increased blood flow to the left lung and a further decrease in Pa,O2. Addition of CO2 to the inspired gas resulted in an increased diversion of blood flow away from the hypoxic lung but a decrease in Pa,O2. Respiratory alkalosis induced by over-ventilation decreased the hypoxic vasoconstriction and increased Pa,O2 slightly. However acid-base changes induced by infusion of 1 N-lactic acid or 8.4% NaHCO3 had no significant effects on the magnitude of the hypoxic vasoconstriction, or on Pa,O2, during hypoxic ventilation of the left lung. The magnitude of hypoxic vasoconstriction and Pa,O2 in the experiments described in 3, 4 and 5 above were positively correlated (r = 0.885), showing that the vasoconstriction may help to maintain Pa,O2. It is suggested that the effects of CO2 on h.p.v. and Pa,O2 may be explained largely by the changes in alveolar oxygen pressure (PA,O2) which are secondary to changes in PA,CO2.