The prognostic value of radionuclide angiocardiography was examined in patients with suspected coronary artery disease. Nine hundred and eight patients who underwent rest and exercise radionuclide angiocardiography without subsequent cardiac catheterization were followed for a median of 4.6 years. Fifty-two cardiovascular deaths and 28 nonfatal myocardial infarctions occurred during the follow-up period. Thirty-nine radionuclide angiocardiographic and clinical variables were analyzed in association with the end points of cardiovascular death, total cardiac events and death from all causes using the Cox proportional hazards model and Kaplan-Meier survival estimates. Univariable analysis identified the exercise ejection fraction as the best predictor of cardiovascular death (chi-square = 82), total cardiac events (chi-square = 84) and death from all causes (chi-square = 66). A small subset of patients (n = 45) with an exercise ejection fraction less than 0.35 were at high risk for future cardiac events, whereas most patients (n = 776) had an exercise ejection fraction greater than or equal to 0.50 and a low probability of a subsequent event. Three variables--the exercise ejection fraction, the exercise change in heart rate, and gender--contained independent prognostic information determined by multivariable analysis. The exercise ejection fraction was the strongest independent predictor (p less than 0.0001) for every end point. The measurement of ventricular function during exercise provides important independent prognostic information in patients with suspected coronary artery disease. Radionuclide angiocardiography successfully identifies patients requiring invasive assessment, and the low probability of cardiac events in patients with good exercise ventricular function obviates the need for interventional therapy.