The factors which could influence the selection of cases for post-mortem examination were analysed during a 6-month period. The variables studied included the age and sex of the patients, the length of terminal hospitalization, the principal disease, the certainty of the clinical diagnostics and the extent of the clinical investigation. Both clinical and autopsy records were reviewed. Seventy-four patients died at the Department of Surgery during this period of whom 50 (68% autopsy rate) were autopsied. The autopsy rate was influenced by (a) the length of the terminal hospitalization, (b) the diagnosis of the principal disease and, to a certain degree, (c) the extent of the clinical investigation. Thus, patients who were hospitalized for a shorter period, had no clinically diagnosed malignant tumours and were not investigated with more sophisticated methods were more frequently autopsied. The discrepancy rate between principal clinical and post-mortem diagnoses was 28% and was not influenced by the use of modern investigative methods.