In a hospital-based cohort of 8903 black and white women, we investigated medical and socioeconomic risk factors that may explain the known increase in premature births among black women. Among the medical conditions examined, only the maternal hematocrit level (or some related factor) explained a substantial proportion (60 percent) of the increased rate of premature births to black women. Four economic, demographic, and behavioral predictors of prematurity were also examined: age less than 20 years, single marital status, receiving welfare support, and not having graduated from high school. The number of these socioeconomic risk factors occurring in a woman was strongly predictive of premature birth of her infant, regardless of the particular risk factors present. The presence of any one factor was associated with a moderate increase in the risk of prematurity (7.0 percent as compared with 4.6 percent with no risk factors present); the presence of two or more characteristics was associated with a much higher risk (11.2 percent). When the number of these four risk factors pertaining to an individual woman was taken into account, race was no longer a significant predictor of premature birth (odds ratio, 1.22; 95 percent confidence interval, 0.94 to 1.59). When both the maternal hematocrit level and the number of the four socioeconomic risk factors were taken into account, essentially all of the racial variation in prematurity was explained, with the odds ratio for prematurity among blacks being 1.03 (95 percent confidence interval, 0.79 to 1.35). We conclude that the racial difference in the rate of premature birth is attributable to specific medical and socioeconomic characteristics.