To test the hypothesis that the recent substantial decline in the United States neonatal mortality rate (20.0/1000 in 1950 to 11.6/1000 in 1975) is associated with improvements in perinatal medical care, we examined this change in relation to the two primary components which determine neonatal mortality: birthweight distribution and birthweight-specific mortality. No improvement in the weight distribution of U.S. live births has occurred during this 25-year period, indicating that the change in neonatal mortality is attributable to improved survival for one or more birthweight groups. Decline in the mortality rate in the first 15 years was slow; three-fourths of the decline in the entire 25-year period occurred since 1965. With the exception of perinatal medical care, factors known to affect survival at a given birthweight have not changed in prevalence in the 25-year period. It is a plausible hypothesis that improved perinatal medical care is a major factor in declining neonatal mortality in the U.S.