Approximately 8 million patients at risk for coronary artery disease undergo noncardiac surgery annually in the United States. This study defined the appropriateness and cost of evaluating these patients with stress imaging tests. Before noncardiac surgery, 178 consecutive patients were prospectively studied by stress imaging. Pretest cardiac risk (low, intermediate, high) was established by interviewing the referring physician and separately by a cardiologist on the basis of the nature of noncardiac surgery and Eagle's clinical criteria. Patients were followed-up for alterations in management and perioperative events until hospital discharge. Referring physicians and cardiologists identified low risk in 24% and 54% of patients, respectively (p < 0.0001). Of 96 patients identified as low risk by cardiologists, 75 had minor surgery and 21 had major surgery, but no clinical risk factors. In the remaining 82 patients with major surgery, ischemia and other severe abnormalities were detected in 19 (23%) patients. At follow-up, no perioperative complications occurred in minor surgery; one patient with major surgery but no clinical risk factors died from complications related to hypertrophic cardiomyopathy. Patients with at least one clinical risk factor undergoing major surgery but who did not have ischemia on stress testing (n = 63) had two complications (infarction and unstable angina). Intervention (revascularization and surgical cancellation) was probably the explanation for the absence of events in 19 patients with ischemia. With a weighted mean Medicare reimbursement ($386), the use of a simple selection algorithm based on noncardiac surgery and clinical risk to avoid testing low-risk patients would have an average cost of $214 per patient, representing a 45% savings.