Objectives. This study was designed to establish the appropriate diagnostic criteria for positive dobutamine electrocardiographic (ECG) stress test results and to compare their accuracy with those of dobutamine two dimensional echocardiography and perfusion scintigraphy. Background. Conventional criteria for positive findings on ECG exercise testing may not be appropriate for use with dobutamine ECG stress testing. Methods. One hundred twenty-nine consecutive patients with an interpretable ECG and without previous myocardial infarction were prospectively studied at the time of coronary arteriography. All completed a standard dobutamine protocol (5 to 40 mu g/kg body weight per min in 3-min dose increments) without side effects. Significant coronary artery disease, defined as >50% lumen diameter stenosis of a major epicardial coronary artery on coronary angiography, was present in 83 patients. Empiric receiver operating curves were generated for various ECG criteria derived from computer-averaged signals. Results. The best ECG criterion, with a sensitivity of 42% and a specificity of 83%, was an ST segment shift, relative to baseline, of 0.5 mm 80 ms after the J point. The sensitivity of this criterion was greater than that of the conventional criterion of 1 mm ST segment depression 60 (23%) or 80 (18%) ms after the J point, was comparable to that of chest pain occurring during the test (44%, p = NS) but remained inferior to the sensitivities of technetium-99m methoxyl isobutyl isonitrile (mibi) perfusion (76%) or stress echocardiography (76%, p < 0.001, for both). The specificity of this criterion was not significantly different from that of technetium-99m mibi perfusion tomography (65%) or stress echo cardiography (89%) but was superior to that of chest pain (59%, p < 0.025). Conclusions. We conclude that this new criterion for dobutamine electrocardiography is specific but that an imaging technique is still required to accurately predict coronary artery disease.