A new method has been developed for measuring virtually continuous distributions of ventilation-perfusion ratios (V(A)/Q) based on the steadystate elimination of six gases of different solubilities. The method is applied here to 12 normal subjects, aged 21-60. In nine, the distributions were compared breathing air and 100% oxygen, while in the remaining three, effects of changes in posture were examined. In four young semirecumbent subjects (ages 21-24) the distributions of blood flow and ventilation with respect to V(A)/Q were virtually log-normal with little dispersion (mean log standard deviations 0.43 and 0.35, respectively). The 95.5% range of both blood flow and ventilation was from V(A)/Q ratios of 0.3-2.1, and there was no intrapulmonary shunt (V(A)/Q of 0). On breathing oxygen, a shunt developed in three of these subjects, the mean value being 0.5% of the cardiac output. The five older subjects (ages 39-60) had broader distributions (mean log standard deviations, 0.76 and 0.44) containing areas with V(A)/Q ratios in the range 0.01-0.1 in three subjects. As for the young subjects, there was no shunt breathing air, but all five developed a shunt breathing oxygen (mean value 3.2%), and in one the value was 10.7%. Postural changes were generally those expected from the known effects of gravity, with more ventilation to high V(A)/Q areas when the subjects were erect than supine. Measurements of the shunt while breathing oxygen, the Bohr CO(2) dead space, and the alveolar-arterial oxygen difference were all consistent with the observed distributions. Since the method involves only a short infusion of dissolved inert gases, sampling of arterial blood and expired gas, and measurement of cardiac output and minute ventilation, we conclude that it is well suited to the investigation of pulmonary gas exchange in man.