In this study transesophageal echocardiography was utilized for detecting air embolism in dogs in the supine position and in patients undergoing neurosurgery in the sitting position. In dogs, the threshold dose of venous air for detection was determined using either a bolus injection or continuous infusion of air via the jugular vein for up to three minutes. The ability to detect air in the aorta also was determined by a bolus injection into the left ventricular via an arterial catheter. For venous injection of air, the threshold dose by bolus was 0.02 ml/kg. When given by infusion, air could be detected in all cases by both contrast echocardiogram and Doppler sound changes at the rate of 0.05 ml . kg-1 . min-1. When air was injected into the left ventricle, the threshold dose was 0.001 ml/kg using contrast echocardiogram. In the clinical evaluation, air was clearly demonstrated in five of six patients by transesophageal echocardiogram along with appropriate changes in Doppler sounds, pulmonary artery pressure, and end-tidal carbon dioxide concentration. Our results suggest that transesophageal echocardiography may be a more sensitive and accurate method for detecting venous air embolism than other commonly used monitors for patients undergoing neurosurgical procedures in the sitting position. This device may also be able to detect air in the aorta in patients experiencing paradoxical air embolism during surgery due to intracardiac or pulmonary shunts.