To assess the usefulness of transesophageal echocardiography in the evaluation of proximal coronary artery stenosis, 111 consecutive patients (mean age 61 years) who had intraoperative transesophageal echocardiography and coronary angiography within 1 week of surgery were studied. Transesophageal echocardiography visualized the entire length of the left main artery (0.2 to 2.2 cm, mean 0.93), 0.2 to 2.2 cm of the proximal left anterior descending artery and 0.1 to 3.4 cm of the proximal left circumflex artery in 103 patients (93%) and 0.1 to 4.6 cm of the proximal right coronary artery in 55 patients (49%). In the coronary artery segments visualized by echocardiography and compared with the corresponding angiographic segments, transesophageal echocardiography correctly identified 23 (96%) of 24 left main stenoses, 11 (78%) of 14 stenoses involving the left anterior descending artery, 6 (75%) of 8 left circumflex stenoses and all 7 stenoses (100%) of the right coronary artery. In all seven patients with ostial stenosis (left main artery in five and right coronary artery in two), the condition was correctly diagnosed by this technique. The sensitivity and specificity of transesophageal echocardiography in the overall evaluation of proximal coronary artery stenosis as customarily defined by angiography were 96% and 99% for the left main artery, 48% and 99% for the left anterior descending artery, 67% and 100% for the left circumflex artery and 37% and 100% for the right coronary artery, respectively. The distance of the stenotic lesion from the origin of the vessel by transesophageal echocardiography also correlated well with that measured by angiography (r = 0.63 to 0.99). Color flow mapping was found to be less useful because significant narrowing of the flow channel was seen in only 40 of 53 stenoses visualized by two-dimensional echocardiography. © 1992, American College of Cardiology Foundation. All rights reserved.