Many previous studies have reported the value of autopsy in assessing clinical diagnostic accuracy. None of them however, assessed the value of autopsies in a specific clinical speciality. The authors reviewed the findings of 123 consecutive neurosurgical autopsies with reference to the premortem clinical diagnoses. The study showed that 7% of cases had a wrong clinical diagnosis and in 9% of cases the clinical diagnosis was incomplete. Only in 5% of all cases knowledge of the autopsy findings would have led to a change in management and outcome. The autopsies also confirmed that 11% of cases died following a surgical complication and in 3% of cases the primary cause of death was non-neurosurgical. The latter was a previously unrecognised finding in 8% of autopsies. The autopsy will remain a valuable means of clinical audit and the increasing financial pressures to reduce the number of autopsies should be resisted.