A comprehensive study was made of the value of necropsies carried out over a six month period in Peterborough Health Authority, using a modification of the format proposed by Schned et al in Vermont, United States of America. Cause of death, clinical correlation, clinical factors contributing to death and clinical evaluation of the necropsy were recorded. The principal diseases were cardiac (33%), and neoplastic (29%). The pathologist disagreed with the clinical cause of death in 13% of cases and major unsuspected diagnoses were found in 30%. Seventy nine per cent of necropsies provided additional information, including feedback on clinical investigations. Clinicians' questions were answered fully in 87%. Necropsies were rated as "very valuable" in 44% of cases, "valuable" in 54%, and "of no value" in only 2%. In 14% of cases the clinical consultant stated that the results of the necropsy would affect future clinical practice. These findings underline the essential place of necropsies in both audit and postgraduate education. The format described could conveniently be used by other pathology departments as part of an audit program.