Data upon all births and infant deaths in New York City in 1968 are analyzed using methods for the analysis of multidimensional contingency tables. These methods provide estimates of the effect of variations in prenatal care upon the relative risks of low birth weight and neonatal and postneonatal mortality, controlling for a wide variety of factors which tend to "select" women into a program of prenatal care. Significant relationships between lack of prenatal care and infant mortality are estimated, but these occur mainly via the relationship of inadequate prenatal care to low birth weight. Furthermore, among white mothers who delivered on a private service, those receiving inadequate levels of prenatal care experienced only slightly increased risks of a low birth weight infant. In contrast, white mothers who delivered on a general service, and all black mothers, experienced substantially increased risks when receiving inadequate prenatal care. A variety of behavioral characteristics of mothers were not controlled in these analyses, and thus clear causal inferences concerning the efficacy of prenatal care cannot be drawn. These analyses do, however, identify a significant population of women at substantial risk.