Previous studies have suggested that maternal smoking is negatively associated with a Down syndrome live birth. We analyzed the data of the U.S. Perinatal Collaborative Study in a search for racial variation in Down syndrome risk factors. There were 22 cases in 25,346 live births to smoking mothers (4/10,780 blacks, 18/13,320 whites, and 0/1,246 other races) and 42/29,130 live births to nonsmoking mothers (24/14,665 blacks, 14/11,694 whites, and 4/2,771 others). The crude overall rates per 1,000 live births were 0.4 in black smokers and 1.6 in black nonsmokers but 1.4 in white smokers and 1.2 in white non-smokers. Adjusted for maternal age, the summary relative risk for a Down syndrome live birth to a smoking mother was 0.2 in blacks (95% interval 0.1-0.7) but 1.2 in whites (95% interval 0.6-2.5). Stratification on variables associated with socioeconomic status or gestational age at time of entry into the study did not alter the racial difference. A comparison of smokers with those who never smoked revealed essentially the same trends. Among all nonsmokers the ratio of the maternal age-adjusted risks for a Down syndrome live birth in whites compared with blacks was 0.7 (95% interval 0.3-1.3), and among all smokers this ratio was 3.6 (95% interval 1.3-9.9). If the results are not attributable to statistical fluctuation or undetected confounding, then differences in the probability of intrauterine survival of the Down syndrome fetus would appear to be one plausible explanation for the difference.