Specificity and sensitivity of exercise ECG were compared to exercise stress thallium 201 scanning and coronary arteriograms in 70 male patients with typical or atypical chest pain complaints, without previous myocardial infarction. 50 patients (group I) did not receive any treatment; 20 patients (group II) received digitalis as preventive treatment of atrial arrythmias or for no particular reason. Only subjects with concordant results in radionuclide and angiography examinations were considered as coronary artery disease patients. Exercise stress tests were performed sitting on the bicycle ergometer using a progressive loading profile (30 W for 3 min), to the symptom-limited capacity (VO<sub>2</sub> SL). Positive exercise ECG were confirmed on ST decrease (1.5 mm) or on absence or increase in R-wave-amplitude modifications (V5). In the group I patients, ECG-ST-modification sensitivity, specificity, predictive value (+) and efficiency were, respectively, 92, 82, 86 and 88%. R-wave-variation sensitivity, specificity, predictive value (+) and efficiency were, respectively, 41, 74, 65 and 56%. In the group II patients, ST-depression sensitivity, specificity, predictive value (+) and efficiency were, respectively, 100, 33, 59 and 65%. R-wave-variation sensitivity, specificity, predictive value (+) and efficiency were, respectively, 50, 70, 63 and 60%. It was concluded that R-wave-amplitude variations induced lower false positive responses than ST-segment depression in patients under digitalis treatment. False negative responses were unfortunately 50% using the R-wave criterion. Erercise ECG was finally judged as a poor indicator of CAD in patients under digitalis treatment.