Six pentobarbital-anesthetized dogs were prepared with endobronchial tubes and electromagnetic flow probes. The effects of changing inspired oxygen concentrations (FIO2 = 1, 0.21, 0.15, 0.1, 0.075, 0.05, and 0) were tested on test segments of different size corresponding to left lower lobe, left upper lobe-lingula, left lung, right lung, right lung plus left lower lobe, right lung plus left upper lobe-lingula, and whole lung. In each test the rest of the lung received oxygen. Hypoxic pulmonary vasoconstriction is demonstrated by both diversion of blood flow away from hypoxic test segments and by increased perfusion pressure. Flow diversion (FD%) decreases with the size of the hypoxic test segment (%QSN) from a maximum of 75% for very small segments to zero when the whole lung is hypoxic. FD% increases linearly as alveolar oxygen tension (PAO2) of the test segment is decreased in the range of 130--28 Torr. When mixed venous oxygen tension (PVO2) is less than 45 Torr FD% is reduced. These relationships are described by FD% = [74.99 - 0.0778 (%QSN) - 0.00661 (%QSN)2] [1.268 - 0.0096 (PAO2)] [0.47 + 0.012 (PVO2)], with r = 0.92 and standard error for prediction of 8.4%. Pulmonary perfusion pressure changes (PPH/PPN) increase with the size of the hypoxic test segments from 0 with very small segments to approximately 2.2 for the hypoxic whole lung. For all test segments PPH/PPN increases linearly with PAO2. These relationships are described by PPH/PPN = 1 + [0.0043 (%QSN) + 0.000072 (%QSN)2] [1.234 - 0.0096 (PAO2)], with r = 0.91 and standard error for prediction of 0.3 units. Responses to hypoxic pulmonary vasoconstriction in dogs are therefore shown to be predictable and continuous, and the physiological basis for action of each of the variables is discussed.