A prospective study of preoperative exercise testing was carried out in 200 patients older than 40 years scheduled for elective major noncardiac surgery under general anesthesia. The exercise test response was electrocardiographically positive in 32 patients (16%) (2 patients had a markedly positive test), equivocal in 11 patients (5.5%) and negative in 157 patients (78.5%). The patients were followed with serial pre- and postoperative electrocardiograms (ECGs) and determinations of serum creatine kinase (CK) and CK-MB. Six patients (3%) had primary endpoints: 3 (1.5%) died postoperatively and 3 (1.5%) had definite postoperative myocardial infarction. Secondary endpoints of suspected postoperative myocardial ischemia/injury diagnosed by ECG or elevation in CK-MB levels occurred in 27 patients (14%). Endpoint events were more common in patients aged 70 years or older. Endpoint events were also more common in patients with an abnormal (positive or equivocal) preoperative exercise test response than in those with a negative response (27% vs 14%); however, preoperative exercise results were not statistically significant independent predictors of cardiac risk. Using multivariate analysis, the only statistically significant independent predictor of risk was the preoperative ECG. Endpoint events were more common in patients with an abnormal than in those with a normal ECG (23% vs 7%, p less than 0.002). Because the results of exercise testing do not appear to add substantially to the risk separation provided by the ECG at rest, exercise testing is not recommended as a routine preoperative method for assessing perioperative risk in older patients who are being evaluated before major elective noncardiac surgery under general anesthesia.