To assess the reliability of the exercise ECG in detecting silent ischemia, ECG results were compared with those of stress-redistribution thallium-201 single-photon emission computed tomography (SPECT) in 116 patients with prior myocardial infarction and in 20 normal subjects used as a control. The LV was divided into 20 segmental images, which were scored blindly on a 5-point scale. The redistribution score was defined as thallium defect score of exercise subtracted by that of redistribution image and was used as a measure of amount of ischemic but viable myocardium. The upper limit of normal redistribution score (= 4.32) was defined as mean +2 standard deviations derived from 20 normal subjects. Of 116 patients, 47 had the redistribution score above the normal range. Twenty-five (53%) of the 47 patients showed positive ECG response. Fourteen (20%) of the 69 patients, who had the normal redistribution score, showed positive ECG response. Thus, the ECG response had a sensitivity of 53% and a specificity of 80% in detecting transient ischemia. Furthermore, the 116 patients were subdivided into 4 groups according to the presence or absence of chest pain and ECG change during exercise. Fourteen patients showed both chest pain and ECG change and all these patients had the redistribution score above the normal range. Twenty-five patients showed ECG change without chest pain and 11 (44%) of the 25 patients had the abnormal redistribution. Three (43%) of 7 patients who showed chest pain without ECG change had the abnormal redistribution score. Of 70 patients who had neither chest pain nor ECG change, 19 (27%) had the redistribution score above the normal range. Thus, limitations exist in detecting silent ischemia by ECG in patients with a prior myocardial infarction, because the ECG response to the exercise test may have a low degree of sensitivity and a high degree of false positive and false negative results in detecting silent ischemia.