Cardiac complications represent a major risk to patients undergoing vascular surgery. Diabetic patients may be particularly prone to such complications due to the high incidence of concomitant coronary artery disease, the severity of which may be clinically unrecognized. Attempts to stratify groups by clinical criteria have been useful but lack the predictive value of currently used noninvasive techniques such as dipyridamole-thallium scintigraphy. One hundred one diabetic patients were evaluated with dipyridamole-thallium scintigraphy before undergoing vascular surgery. The incidence of thallium abnormalities was high (80%) and did not correlate with clinical markers of coronary disease. Even in a subgroup of patients with no overt clinical evidence of underlying heart disease, thallium abnormalities were present in 59%. Cardiovascular complications, however, occurred in only 11% of all patients. Statistically significant prediction of risk was not achieved with simple assessment of thallium results as normal or abnormal. Quantification of total number of reversible defects, as well as assessment of ischemia in the distribution of the left anterior descending coronary artery was required for optimum predictive accuracy. The prevalence of dipyridamole-thallium abnormalities in a diabetic population is much higher than that reported in nondiabetic patients and cannot be predicted by usual clinical indicators of heart disease. In addition, cardiovascular risk of vascular surgery can be optimally assessed by quantitative analysis of dipyridamole-thallium scintigraphy and identification of high- and low-risk subgroups.