To clarify the mechanism of ST-T changes in electrocardiogram, these changes that appear on the direct unipolar electrogram recorded from the ventricular surface were compared with the changes of transmembrane potentials simultaneously recorded from the vicinity of direct unipolar electrogram. This study was conducted on the in situ canine heart under conditions of total and local hypoxemia. Method Seventy-seven mongrel dogs weighing about 15 kg were used, and each animal was anesthetized intravenously by pentobarbital sodium 25 mg per kg of body weight. Then the heart was exposed after performing thoracotomy under artificial positive pressure respiration. In order to prevent the undue movement of recording site on ventricular surface, the inner and outer rings attached to the supporting frame were placed over the pericardial surface, and then the portion of the pericardium within the inner ring was dissected away. Ringer's solution with the temperature of approximately 33°C was constantly dripped upon the epicardial surface, in order to prevent dryness. Using the central terminal system of Wilson, a unipolar surface electrogram was taken simultaneously with transmembrane potentials from the epicardium within the inner ring. The surface electrode consisted of tungsten wire 100 micron in diameter, and was gently placed on the ventricular surface by micromanipulator. The transmembrane potential of the individual cell was recorded from a flexibly mounted ultramicroelectrode such as described by WOODBURY and BRADY (1956). The indifferent electrode consisted of tungsten wire with a diameter of about 100 micron, placed 2 to 4 mm away from the site of penetration by microelectrode. Total hypoxemia of the ventricles was induced by letting a dog inhale the mixture-gas of oxygen 4% and of nitrogen gas 96% for from 6 to 8 minutes. For producing the local hypoxemia of the ventricle, approximately 120 ml of venous blood was taken from the femoral vein by syringe. At this time special care was paid to venous blood, that is, venous blood was kept from exposure to room-air by a layer of floating liquid paraffin in the syringe. This venous blood was kept at approximately 36°C. This venous blood was perfused at a pressure of 100 mmHg into the artificial coronary circuit made between the left carotid artery and the anterior descending branch of the canine left coronary artery (heparinized 500 units per kg/body weight). The room temperature was kept between 24 to 25°C through-out the experiment.