Objective: To determine how often a perinatal autopsy identified the cause of death, and how frequently this information changed recurrence risk estimates or altered parental counseling. Methods: We reviewed all autopsies of fetal stillbirths and briefly viable neonates performed by one perinatal pathologist at the University of Alabama Hospital from 1992 to 1994. Results: Four hundred sixteen fetal and early neonatal deaths occurred at our hospital from January 1, 1992, to June 1, 1994. Consent for an autopsy examination was granted for 139 of these (33%), and all autopsies were performed by a single perinatal pathologist. Abnormalities likely to be the cause of death were identified in 130 of 139 cases (94%). Ninety-one subjects did not have structural anomalies: In 14 cases autopsy revealed previously unsuspected pathology that altered parental counseling; in 68 cases autopsy findings were consistent with the clinical obstetrical diagnosis; and in nine cases the cause of death could not be identified. Forty-eight subjects were anomalous. Thirty-seven of these (79%) had been evaluated by antenatal ultrasonography, and autopsy identified additional abnormalities in 51% (19 of 37). In the 11 deaths evaluated neonatally, a previously unsuspected diagnosis likely to be the cause of death was identified in three. Overall, autopsy findings changed recurrence risk estimates and/or parental counseling in 36 of 139 cases (26%). Conclusion: The cause of fetal or perinatal death was determined by autopsy in 94% of cases in our series. Counseling and recurrence risk estimates were altered by autopsy findings in 26%.