The Rural Infant Care Program (RICP), initiated in 1979, was developed to improve perinatal health care in ten rural sites with histories of high infant mortality rates. Time-series regression models indicate that neonatal mortality rates were reduced, following program initiation, by 2.6 per 1,000 live births (p = .0002); black neonatal mortality rates were reduced by an estimated 4.5 per 1,000 (p = .0004). Three sets of comparison areas exhibited no significant changes in rates. Postneonatal mortality rates did not increase in the target areas following initiation of RICP, indicating that deaths were not merely being postponed. Nine of ten individual sites showed reductions in infant mortality following program initiation. Birthweight-specific mortality data indicated that the decline was due mainly to reductions in neonatal mortality among low-birthweight infants. No reductions in the incidence of low birthweight were observed in the target areas. Substantial gaps in the delivery of prenatal care remained due to the continuing poverty of the population and the resultant lack of financial coverage for health services. We conclude that improved perinatal medical care can reduce infant mortality in poor rural areas to average levels experienced in the United States, and that the high rates still observed in some rural counties are unnecessary.