The incidence and the outcome of systemic air embolism in 340 consecutive patients who underwent cardiac surgery under cardiopulmonary bypass in this unit for congenital defects of the cardiac septa and diseases involving the aortic and mitral valves have been studied. This was thought to have occurred in 40 patients, of whom 10 died. The distribution of air embolism according to the types of operation undertaken was as follows: six of 127 for atrial septal defect; six of 36 for ventricular septal defect; seven of 42 for mitral valve replacement; seven of 47 for aortic valve debridement; and 14 of 55 for aortic valve replacement. The cause was considered to have been systolic ejection of air into the aorta which, following cardiotomv. had been trapped in the pulmonary veins, the left atrium, the ventricular trabeculae, and the aortic root. Since the adoption of a more rigid 'debubbling' routine, air embolism has not occurred. The incidence of pulmonary complications occurring in these patients after bypass was studied. Unilateral atelectasis, which occurred in five patients, resulted from retained bronchial secretions in all and was cured by bronchoscopic aspiration in all. The cause of bilateral atelectases, occurring in nine patients and fatal in eight of these, appeared to be related to cardiopulmonary factors and not to air embolism. Acute air injection made into the pulmonary artery of a dog resulted in pulmonary hypertension and a grossly deficient pulmonary circulation, but changes were largely resolved within a week. In view of this, it is considered that pulmonary air embolism may temporarily embarrass the right heart after the repair of a ventricular septal defect in a patient with an elevated pulmonary vascular resistance and diminished pulmonary vascular bed.An appraisal of the part played by arterial air embolism in complications following open-heart surgery and of safeguards for preventing this has been attempted. HISTORICAL SYNOPSIS Brandes (1912) showed by means of radiography and by histological sections that the sudden death of his patient during the injection of bismuth paste into an empyema cavity was due to the paste passing into the pulmonary veins and the heart, from which it was pumped into the coronary and cerebral arteries.In a scholarly review, Schlaepfer (1922) discussed factors influencing the distribution of air to the cerebral arteries (readily recognized from the spider-web pattern visible in the retina) and to the coronary arteries (indicated by ischaemic changes on the electrocardiogram). He observed that air injected into the pulmonary veins remained in the atrial appendix when the animal was placed in the Trendelenburg position, but that it passed into the ventricle and was ejected into the right coronary artery and into the cerebral arteries on raising the body to the supine position. He recommended that pneumothorax be induced in the recumbent position and that the trunk be depressed immediately should embolism occur. He showed that air in the pulmonary artery did not tra...