A B S T R A C T Multiple indicator dilution studies of the pulmonary circulation were carried out in conscious, resting and exercising, and anesthetized dogs under conditions where there was no pulmonary edema. Labeled red cells, water, and albumin were injected together into the pulmonary artery, and effluent dilution patterns were obtained from the descending thoracic aorta. The product of the mean transit time differences between labeled water and red cells, and the pulmonary water flow was used to estimate extravascular parenchymatous water; and this was expressed as a proportion of the water content of the blood-drained lung at postmortem examination. These estimates of the proportional water content were found to increase with flow, and to approach an asymptotic value. Reconsideration of the flow patterns in capillaries, however, led to the postulate that extravascular water should be calculated, utilizing as the appropriate vascular reference a substance that uniformly labels the water in red cells and plasma, and which is confined to the circulation, rather than a tracer that only labels red cells. The mean transit time of this substance is approximated by the sum of the mean transit times of labeled red cells and albumin, each weighted according to the proportion of the water content of blood present in that phase. The values for lung water content so computed also increased with flow, and appeared to approach an asymptote that corresponded to approximately two-thirds of the wet lung weight. The estimated values for the water space after pentobarbital anesthesia corresponded to the lower values obtained in the resting conscious animals. When the anesthetized animals were also bled, the estimated water space was disproportionately large, in relation to the previous values. These experimental results support the hypothesis that dilutional This work was presented in part at the 51st Annual