Using data from the Missouri maternally linked files (1989-1997), the authors examined the association among maternal obesity, obesity subtypes, and spontaneous and medically indicated preterm (<37 weeks) and very preterm (<33 weeks) births in singletons and twins. Adjusted odds ratios were obtained with correction for intracluster correlation. The prevalence of obesity increased by 77% over the study period (p(trend) < 0.001). Obese mothers had a lower risk for spontaneous preterm birth, and this was more pronounced among twins (odds ratio = 0.68, 95% confidence interval: 0.62, 0.75) than singletons (odds ratio = 0.84, 95% confidence interval: 0.82, 0.87). However, this association was present only among obese women who gained less than 0.69 kg/week for singletons and between 0.23 and 0.69 kg/week for twins. By contrast, obese mothers with singleton gestation had about 50% greater odds of medically indicated preterm (odds ratio = 1.46, 95% confidence interval: 1.39, 1.54) and very preterm (odds ratio = 1.49, 95% confidence interval: 1.34, 1.65) births, and the risk increases with ascending severity of obesity (p(trend) < 0.01). For extreme obesity, the risk of medically indicated preterm and very preterm births was almost double that for nonobese women. Similar findings were observed in twins. These data suggest that obesity increases the risk for medically indicated but not spontaneous preterm birth in both singletons and twins.