To evaluate the accuracy of exercise two-dimensional echocardiography for the recognition of coronary artery disease, 53 patients (46 men and 7 women, age range 35 to 69 years) without either previous myocardial infarction or resting wall motion abnormalities, were studied. According to coronary angiography 26 had normal coronary arteries, 14 had one-vessel, seven had two-vessel, and six had three-vessel disease. After withdrawal of any therapy, all patients underwent a single exercise stress test with a stress table during which cine-loop digitized echocardiography was acquired and 74 MBq of thallium-201 (TI-201) were injected. Echocardiographic images were evaluated at rest and at peak exercise. Three-view planar scintigraphic images were collected immediately after exercise and 4 hours later. For the overall recognition of coronary artery disease, exercise electrocardiography had 77.8% sensitivity and 65.4% specificity; myocardial scintigraphy had 100% sensitivity and 92.3% specificity; and exercise echocardiography had 92.6% sensitivity and 96.2% specificity (both NS versus myocardial scintigraphy). Global accuracy was 71.7% for exercise electrocardiography, 94.3% for stress echocardiography, and 96.2% for myocardial scintigraphy. For the classification of the individual involved coronary arteries, the sensitivity of myocardial scintigraphy was 84.8% and that of exercise echocardiography was 63% (p less than 0.01); the related specificities were 98% and 98.2% respectively (NS). It may be concluded that exercise echocardiography is highly accurate for the recognition of coronary artery disease, whereas it appears less sensitive in the identification of the involved vessels, particularly in patients with multivessel disease.