We undertook a cross-sectional epidemiologic study of potential determinants of parental consent for neonatal autopsy at the Brigham and Women's Hospital, Boston. Data were abstracted from maternal and infant medical records in 184 cases of neonatal death occurring between January 1982 and October 1984. The overall consent rate for neonatal autopsy was 72%. Multivariate analysis by logistic regression found previous fetal loss, gestational age, and cause of death to be significantly different between the groups of consenters and nonconsenters. Parents least likely to consent to autopsy were those who had no history of previous fetal loss, who had pregnancies in which the birth weight of the infant was less than 1000 g or the gestational age was less than 28 weeks, or those who had an infant die of extreme prematurity. Factors not significantly associated with consent were maternal age, race, marital status, transfer status, type of prenatal care, the infant's sex, and the staff position of the requester. A second phase of the study surveyed physicians' attitudes regarding the importance of neonatal autopsy. The staff position and previous experience of the physician-requester, in addition to the presumed cause of the infant's death, were significantly associated with the rating assigned to the importance of the autopsy. These findings suggest that the mother's past and present obstetrical experience, the presence of extreme prematurity, and possibly the attitude and experience of the physician requesting autopsy permission may exert important influences on the likelihood of obtaining consent for a neonatal autopsy.