An isolated perfused canine lung preparation in which determinants of vascular caliber could be individually controlled was developed. The relation of pulmonary arterial (Pa), venous (PV), and alveolar (PA) pressures was such that Pa greater than PA greater than PV throughout the whole lung. The addition of isoprenaline to the perfusate abolished vascular reactivity. Once stability was reached, vascular cross-sectional area remained acceptably constant for 2.25 hours as judged by normalized conductance. The influence of perfusate hematocrit, blood gas tensions, and pH on pressure-flow relations was then studied in 15 isolated canine lungs. The hematocrit-vascular conductance relation was derived at constant perfusion pressure. Conductance varied linearly with hematocrit over a range of 16.5 to 89.5%. Mean pulmonary arterial blood gas tensions were: PO2 = 121 mm Hg, PCO2 = 28 mm Hg, and pH = 7.46. Acute respiratory acidosis (PO2 = 30 mm Hg, PCO2 = 81 mm Hg, pH = 7.17) and lactic acidosis and hypoxemia (PO2 = 32 mm Hg, PCO2 = 21 mm Hg, pH = 6.96) did not significantly alter this relation. Transformation of the conductance-hematocrit data indicated that hematocrit was the most important determinant of relative apparent viscosity of the blood. Both acute respiratory and lactic acidosis failed to significantly increase relative viscosity within the range of hematocrit usually found in secondary polycythemia.