Sixty-six patients unable to complete a standard preoperative exercise test because of physical limitations were studied to determine the predictive value of individual clinical parameters, of clinical scoring systems based on multifactorial analysis, and of dipyridamole-thallium imaging before major general and vascular surgery. Study endpoints were limited to postoperative myocardial infarction or cardiac death before hospital discharge. There were nine postoperative cardiac events (seven deaths and two nonfatal infarctions). There was no statistical correlation between cardiac events and preoperative clinical descriptors, including individual clinical parameters, the Dripps-American Surgical Association score, the Goldman Cardiac Risk Index score, the Detsky Modified Cardiac Risk Index score, Eagle's clinical markers of low surgical risk, and the probability of postoperative events as determined by Cooperman's equation. There were no cardiac events in 30 patients with normal dipyridamole-thallium scans or in nine patients with fixed myocardial perfusion defects. Of 21 patients with reversible perfusion defects who underwent surgery, nine had a postoperative cardiac event (sensitivity, 100%; specificity, 43%). In the six other patients with reversible defects, preoperative angiography showed severe coronary disease or cardiomyopathy. Thus in patients unable to complete a standard exercise stress test, postoperative outcome cannot be predicted clinically before major general and vascular surgery, whereas dipyridamole-thallium imaging successfully identified all patients who sustained a postoperative cardiac event.