BACKGROUND Recent studies suggest elevated exposure to drinking water disinfection by-products (DBPs) may be associated with decreased risk of preterm birth. We examined this association for exposure to total trihalomethanes (TTHMs), 5 haloacetic acids (HAA5), and total organic halides. METHODS Analysis included 2039 women in a prospective pregnancy study conducted from 2000 through 2004 in 3 study sites. Water samples were collected and analyzed for DBP concentrations. Participant data were collected through interviews, an early ultrasound, and birth records. We assessed the associations between DBPs and preterm birth (<37-weeks' gestation) using log-binomial regression. Discrete-time hazard analysis was used to model the conditional odds of delivery each week in relation to DBP exposure. RESULTS Average second trimester DBP levels were associated with lower risk of preterm birth. Adjusted risk ratios for TTHM levels of 33.1-55.0, 55.1-66.3, 66.4-74.8, and 74.9-108.8 microg/L versus 2.2-4.6 microg/L were 0.8 (95% confidence intervals = 0.5-1.3), 0.9 (0.6-1.4), 0.7 (0.4-1.1), and 0.5 (0.3-0.9), respectively. Risk ratios for HAA5 levels of 17.9-22.0, 22.1-31.5, 31.6-40.4, and 40.5-52.8 microg/L versus 0-0.9 microg/L were 1.1 (0.8-1.7), 0.8 (0.5-1.2), 0.5 (0.3-0.8), and 0.7 (0.4-1.1), respectively. The conditional odds of delivery each week were decreased for the highest TTHM and HAA5 exposure groups versus the low exposure group for gestational weeks 33-40. CONCLUSIONS The probability of preterm birth was not increased with high DBP exposure.