Dipyridamole single-photon emission computed tomography (SPECT) has a high negative predictive value for perioperative cardiac events, but events are infrequent in patients with a positive test. In contrast, dipyridamole echocardiography is more selective for detection of multivessel disease and thus may have a greater specificity for cardiac events. We therefore compared the ability of dipyridamole SPECT and echocardiography to predict perioperative and long-term cardiac events in 133 patients referred for vascular surgery. The group was also evaluated based on clinical features and ejection fraction. Four patients had surgery cancelled because of high risk and were excluded from further analysis. Among the 129 remaining patients, 21 had coronary revascularization (n=12) or an early cardiac end point (n=9). The sensitivity of SPECT for the prediction of early events (90%) was not significantly different from that of echocardiography (66%, p=NS). The specificity of SPECT (68%) was less than that of echocardiography (88%, p <0.001%), as was the accuracy (72% vs 84%, p=0.02). These findings were replicated after exclusion of patients with treatment end points. During long-term follow-up, 12 patients experienced > or = 1 event: 6 died from cardiac causes, 4 underwent revascularization, and 3 had myocardial infarction. Thus, the specificity of SPECT and echocardiography for late events were 58% and 80%, respectively (p <0.001). The 3-year survival of patients without ischemia during echocardiography or at SPECT was not different (93% vs 94%, p=NS).