To evaluate the utility of exercise radionuclide angiography (RNA) and electrocardiography in assessing the severity of coronary artery disease (CAD), 185 patients undergoing coronary angiography were studied prospectively. To avoid work-up bias and to provide an appropriate control group, all patients were simultaneously scheduled for exercise RNA and electrocardiography and for coronary arteriography. All test results were interpreted blinded to other data. Of multiple exercise variables analyzed by stepwise linear discriminant analysis, the independent predictors of disease severity were exercise ejection fraction, ST segment change, and maximum heart rate. These three variables were used to create a set of four equations that determine probabilities of zero, one-, two-, or three-vessel disease (VD). The noninvasive estimate of number of VD in each patient was compared to the angiographic result. Patients without significant CAD were classified correctly 71% of the time, while those with three-VD were predicted correctly in 80%. Fully 90% of patients with predicted three-VD had two- or 3-VD. Conversely, 84% of patients predicted to have zero VD had zero or one-VD. Thus the combined use of exercise RNA and ECG data permits assessment of the presence and severity of CAD.