Fetal deaths, in contrast to infant deaths, have been subject to epidemiologic analysis infrequently. We characterized 574 Massachusetts resident fetal deaths from 1982 and assessed the accuracy of cause-of-death information available from vital records compared with that from corresponding fetal autopsies. The fetal death rate exceeded the neonatal mortality rate. Fetal mortality was higher among black, unmarried, and older mothers. Fetuses of multiple-gestation pregnancies had an unusually high risk of fetal death. Autopsy reports were obtained for 61% of fetal deaths. The underlying cause of death from the fetal death record differed from that on the autopsy report in 55% of cases. Systematic collection of population-based autopsy data is a useful approach for improving the quality and accuracy of mortality statistics on fetal deaths. Many stillbirths remain unexplained, however, and research is needed to identify pathological markers that might reduce the heterogeneity within the fetal deaths currently ascribed to unknown causes.