OBJECTIVE: To determine our knowledge in terms of cause of death and quality of death certification about patients who die in the emergency department. To establish the role of autopsy in this matter. METHODS: Retrospective chart review of all patients dying in an academic emergency department (ED) of a tertiary hospital over a period of 1 year. RESULTS: One hundred and ninety-six patients died in the ED in 1998. In 141/196 patients the cause of death could be determined on clinical grounds. In 53/196 patients, the antemortem clinical diagnosis was unknown. Twenty-nine out of 53 patients underwent autopsy. In all but one patient autopsy revealed the cause of death. After retrospective analysis of all patient data (notes, biology, radiology and pathological investigation), the major causes of death were cardiac (19.4%), cerebral (non-traumatic) (16.8%), trauma (15.3%) and unknown (13.3%). In the patient group with sudden cardiac arrest of unclear origin, the postmortem cause of death was identified as cardiac (51.7%), non-traumatic bleeding (10.3%), infectious (10.3%) and pulmonary embolism (3.4%). In the group of patients with a clinically clear cause of death, who underwent autopsy, 14 class II findings according to the Goldman's classification of autopsy diagnoses (i.e. major diagnosis whose detection would not have altered therapy nor outcome) were found in 16 patients. No class I findings (i.e. major diagnosis whose detection would have altered therapy or outcome) were noted. Altogether, major discrepancies between the antemortem presumed cause of death according to the notes and the real cause of death was found in 15.3%. CONCLUSIONS: Autopsy remains a very important tool to establish the cause of death in patients dying in ED. The concordance between the antemortem presumed cause of death recorded in the patient notes and the real cause (all patient data) is poor.