The question whether an autopsy could have been omitted without adversely affecting diagnostic accuracy in cases in which an unnatural death was not suspected, was investigated by evaluating data from the files of cases handled in 1977 at the Government Department of Forensic Medicine in Uppsala in which the police had requested a simple medicolegal examination. The police report and the result of an external examination were evaluated blindly and cases divided into three groups. In 95 cases (12%) no autopsy, in 360 cases (45%) a partial autopsy, and in 346 cases (43%) a complete autopsy was considered necessary. In three of the 95 cases and in four of the 360 cases a wrong diagnosis was made. In 59 cases (7%) a complete medicolegal examination was performed. In 21 of the latter the circumstances of death were obscure and in 11 of these an unnatural cause of death was found. The investigation confirms that all cases must be handled by skilled medicolegal experts, that the extent of the medicolegal examination can well be assessed by the physician in charge, and that the autopsy procedure can be simplified in many cases and omitted in some, especially in a situation where resources are limited, without significantly affecting diagnostic accuracy in the individual case. However this policy adversely affects scientific studies on larger autopsy material.