OBJECTIVE: In recent years, gains in neonatal survival have been most evident among very low birth weight, preterm, and low birth weight (LBW) infants. Most of the improvement in neonatal survival since the early 1980s seems to be the consequence of decreasing birth weight-specific mortality rates, which occurred during a period of increasing preterm and LBW rates. Although the decline in neonatal mortality has been widely publicized in the United States, research suggests that clinicians may still underestimate the chances of survival of an infant who is born too early or too small and may overestimate the eventuality of serious disability. So that clinicians may have current and needed ethnic- and race-specific estimates of the "chances" of early survival for newborn infants, we examined birth weight/gestational age-specific neonatal mortality rates for the 3 largest ethnic/racial groups in the United States: non-Hispanic whites, Hispanics, and non-Hispanic blacks. Marked racial variation in birth weight and gestational age-specific mortality has long been recognized, and growing concerns have been raised about ongoing and increasing racial disparities in pregnancy outcomes. Our purpose for this investigation was to provide an up-to-date national reference for birth weight/gestational age-specific neonatal mortality rates for use by clinicians in care decision making and discussions with parents. METHODS: The National Center for Health Statistics linked live birth-infant death cohort files for 1995-1997 were used for this study. Singleton live births to US resident mothers with a reported maternal ethnicity/race of non-Hispanic white, non-Hispanic black, or Hispanic (n = 10 610 715) were selected for analysis. Birth weight/gestational age-specific neonatal mortality rates were calculated using 250 g/2-week intervals for each ethnic/racial group. RESULTS: The overall neonatal mortality rates for whites, Hispanics, and blacks were 3.24, 3.45, and 8.16 neonatal deaths per 1000 live births, and the proportion of births <28 weeks was 0.35%, 0.45%, and 1.39%, respectively. Newborns who weighed <1500 g comprised <2.5% of all births in each racial/ethnic group but accounted for >50% of neonatal deaths. For whites, Hispanics, and blacks, >50% of newborns 24 to 25 weeks of gestational age survived. For most combinations of birth weights <3500 g and gestational ages of <37 weeks, the neonatal mortality rate was lowest among blacks, compared with whites or Hispanics. At these same gestational age/birth weight combinations, Hispanics have slightly lower mortality rates than whites. For combinations of birth weights >3500 g and gestational ages of 37 to 41 weeks, Hispanics had the lowest neonatal mortality rate. In these birth weight/gestational age combinations, where approximately two thirds of births occur, blacks had the highest neonatal mortality rate. CONCLUSIONS: Compared with earlier reports, these data suggest that a substantial improvement in birth weight/gestational age-specific neonatal mortality has occurred in the United States. Regardless of ethnicity/race, the risk of a neonatal death does not exceed 50% (the suggested definition for the limit of viability), except for birth weights below 500 g and gestational ages <24 weeks. Notwithstanding, ethnic/racial variations in neonatal mortality rates continue to persist, both in overall rates and within birth weight/gestational age categories. Blacks continue to have higher proportions for preterm and LBW births, compared with either whites or Hispanics. At the same time, blacks experience lower risks of neonatal mortality for preterm and LBW infants, while having higher risks of mortality among term, postterm, normal birth weight, and macrosomic births.