We studied the ability to predict depressed left ventricular ejection fraction (LVEF) from clinical and exercise test findings prior to surgery in consecutive patients who underwent coronary artery bypass grafting (CABG) from 1988 to 1991 (n = 663). Multivariate analysis showed a history of myocardial infarction, pathological Q-wave in resting ECG, systolic blood pressure at maximal exercise and the degree of mitral regurgitation as significant independent predictors of impaired LVEF. The relative risk (RR) of depressed LVEF was markedly increased for a previous history of myocardial infarction (RR 3.3, p < 0.0001) and a pathological Q-wave in resting ECG (RR 2.4, p < 0.0001). All associations found between depressed LVEF and exercise test results were poor, and of little value for discriminating patients with depressed LVEF. Thus, clinical data appear to be better markers of low LVEF than the information obtained from the exercise test.