BACKGROUND Less than one third of patients with fatal pulmonary embolism (PE) are identified prior to autopsy. OBJECTIVE To determine whether the clinical syndromes of acute PE are effective at identifying patients who die of this condition. METHODS Seven hundred seventy-eight autopsy reports at the Buffalo General Hospital from 1991 to 1996 inclusive were reviewed. Inpatient medical records of 67 patients who were identified as having PE as the primary or major cause of death then were analyzed. RESULTS Thirty patients (45% [95% confidence interval, 33 to 57%]) had received a diagnosis of PE prior to death, which was marginally higher than the number previously reported (p < 0.05). The diagnosis of PE was significantly lower (13%; p < 0.01) in patients with COPD or coronary artery disease (33%; p < 0.01). In contrast to the prospective investigation of PE diagnosis data, only a minority of patients (6%) presented with pleuritic pain or hemoptysis, while a significantly larger proportion (24%; p < 0.01) of our patients experienced circulatory collapse. Only 55% were identified as having PE from the following clinical syndromes of PE: isolated dyspnea; pleuritic pain and/or hemoptysis; and circulatory collapse. Among the 30 patients suspected of having PE, only 14 (47%) received IV heparin in therapeutic doses, despite clinical suspicion. CONCLUSIONS Our results show a modest increase in the correct antemortem diagnosis of fatal PE. The current clinical syndromes used as markers for suspecting PE are not sufficient to detect patients who ultimately die of PE. Physicians should maintain a higher index of suspicion since fatal PE does not always present as one of the three clinical syndromes of PE. Once PE is suspected, heparin therapy should be started early.