The ability to stratify cardiac risk before nonvascular surgery using clinical markers and dipyridamole-thallium scanning (DTS) was assessed for patients with known or suspected coronary artery disease unable to exercise. Of 100 consecutively studied patients who proceeded to nonvascular surgery, 9 (9%) experienced greater than or equal to 1 perioperative cardiac ischemic event, including death in 2 patients (2%) and nonfatal myocardial infarction in 2 (2%). Logistic regression identified 2 clinical predictors (age greater than 70 years and history of heart failure), and 1 DTS (thallium redistribution) predictor of events. Of 45 patients with neither clinical variable, none (0%; 95% confidence intervals [CI] 0 to 8%) had events. Of 55 patients with greater than or equal to 1 clinical marker, 9 (16.4%; 95% CI 7 to 26%) had events. Within this subgroup, 1 of 31 patients (3.2%; 95% CI 0 to 16%) without thallium redistribution had events compared with 8 of 24 (33.3%; 95% CI 14 to 52%) with redistribution. An algorithm combining 5 independent clinical and 2 DTS predictors, derived previously in vascular surgery patients, was validated in the 100 nonvascular surgery patients. It is concluded that preoperative planar DTS is most useful to stratify selected nonvascular surgery patients at intermediate or high risk by clinical assessment. However, for almost half of those patients with known or suspected coronary artery disease, DTS may be unnecessary because of sufficiently low predictive value based on simple clinical descriptors.