The association between sudden infant death syndrome (SIDS) and maternal smoking was compared between the United States and Sweden-two countries with different health care and social support programs and degrees of sociocultural heterogeneity. For 1990-1991 among the five US race/ethnic groups studied, SIDS rates ranged from a high of 3.0 infant deaths per 1,000 live births for American Indians to a low of 0.8 for Hispanics and Asian and Pacific Islanders. The SIDS rate for Sweden (using 1983-1992 data) was 0.9. The strong association between maternal smoking and SIDS persisted after controlling for maternal age and live birth order. Adjusted odds ratios ranged from 1.6 to 2.5 for mothers who smoked 1-9 cigarettes per day during pregnancy (compared with nonsmokers) and from 2.3 to 3.8 for mothers who smoked 10 or more cigarettes per day during pregnancy. Although birth weight had a strong independent effect on SIDS, the addition of birth weight to the models lowered the odds ratios for maternal smoking only slightly, suggesting that the effect of smoking on SIDS is not mediated through birth weight. SIDS rates increased with the amount smoked for all US race/ethnic groups and for Sweden. Smoking is one of the most important preventable risk factors for SIDS, and smoking prevention/intervention programs have the potential to substantially lower SIDS rates in the United States and Sweden and presumably elsewhere as well.