The charts of 61 patients treated from 1970 through 1981 were reviewed to determine the clinical outcome after treatment of air embolism from blunt (15 patients) and penetrating (21 gunshot and 25 stabbing) thoracic injuries. The diagnosis of air embolism was confirmed by the presence of air in the coronary arteries (57%), air aspirated from the heart (30%) or major artery (10%), or Doppler findings (3%). All patients were in shock or cardiac arrest, and in 36% of these patients there were early signs of hemoptysis or unexpected arrest after intubation and positive-pressure ventilation. Successful management included (1) early thoracotomy for diagnosis as well as for specific treatment, (2) hilar cross-clamping for control of bronchovenous communication, (3) maintenance of perfusion pressures with fluids, vasopressors, or aortic cross-clamping, and (4) prompt correction of the embolic source, usually a lung resection. The overall survival rate was 44%, which correlated with the mechanism of injury, with associated nonthoracic injuries, and with the occurrence of arrest in a controlled setting. We conclude that (1) air embolism can insidiously occur even in blunt trauma; (2) suspicion should be high with hemoptysis or unexpected arrest; and (3) successful treatment includes immediate thoracotomy for diagnosis, resuscitation, and prompt control of the bronchovenous communication.