Despite the high reported accuracy of exercise echocardiography in the detection of coronary artery disease, factors that compromise its sensitivity and specificity are less clear. This study examined the results of 179 post-treadmill stress echocardiograms in 150 consecutive patients who also underwent cardiac catheterization and in 29 normal persons at low risk for coronary artery disease. Of 114 patients who had significant coronary stenoses at angiography, 96 had an abnormal exercise echocardiogram (overall sensitivity 84%). False negative results correlated with the performance of submaximal exercise, single-vessel disease and moderate (50% to 70% diameter) stenoses. After the exclusion of seven patients performing submaximal exercise, the sensitivity was 90%. In 54 patients without previous infarction performing maximal exercise, the sensitivity was 87%, higher in patients with multivessel coronary disease (96%) than in those with single-vessel disease (79%). After the exclusion of patients with nondiagnostic results, due either to the performance of submaximal stress or the presence of electrocardiographic (ECG) changes at rest, exercise echocardiography had a higher sensitivity than did exercise electrocardiography (87% vs. 63%, p = 0.01). In 36 patients without significant coronary disease, exercise echocardiography had an overall specificity of 86%. After the exclusion of patients with a nondiagnostic test, exercise echocardiography had a specificity of 82% compared with 74% specificity for exercise electrocardiography (p = NS). Similarly, of the 29 normal subjects, 93% had a normal exercise echocardiogram and 97% had a normal exercise ECG (p = NS). Similarly, of the 29 normal subjects, 93% had a normal exercise echocardiogram and 97% had a normal exercise ECG (p = NS). Age, gender, body weight and image quality did not significantly influence the accuracy of exercise echocardiography.(ABSTRACT TRUNCATED AT 250 WORDS)