Reported experience with ventricular assist devices (VAD) routinely includes the rate of thromboembolic events, which is commonly calculated from clinically evident findings. Fifty-four patients have had postcardiotomy circulatory support with the Sarns centrifugal device at our institution. We have reviewed 43 patients who failed to survive VAD support to compare the thromboembolism rate diagnosed clinically to that determined at autopsy. In the 35 patients who had no autopsy, there was one clinically apparent thromboembolic event (2.3%). In eight similar patients who had autopsy, there was no clinically apparent thromboembolism. Five of these eight patients (63%) had acute thromboembolic infarcts determined at autopsy. Three had evidence of pulmonary thromboembolism, two cerebrovascular infarction, two liver infarcts, two splenic infarcts, two kidney infarcts, and one each gastric, pancreatic, prostate, adrenal, cervical, and ileal infarcts. All had left and/or right ventricular infarctions. It is concluded that patients dying following VAD have commonly suffered perioperative myocardial infarction. When evaluating complications associated with VAD, one should consider that the true incidence of thromboembolic events is underestimated by clinical findings.