Exposure during pregnancy to disinfection by-products in drinking water has been hypothesized to lead to several adverse reproductive outcomes. We performed a retrospective cohort study to examine the relation of trihalomethane exposure during the third trimester of pregnancy to low birthweight, term low birthweight, and preterm delivery. We matched Colorado birth certificates from January 1, 1990, through December 31, 1993, to historical water sample data with respect to time and location of maternal residence based on census block groups. After excluding births from all census block groups with no trihalomethane sample data and restricting to singleton white births with 28-42 weeks of completed gestation (>400 gm), we studied 1,893 livebirths within 28 census block groups. We found a weak association of trihalomethane exposure during the third trimester with low birthweight (odds ratio = 2.1 for the highest exposure level; 95% confidence interval = 1.0-4.8); a large increase in risk for term low birthweight at the highest level of exposure (odds ratio = 5.9; 95% confidence interval = 2.0-17.0); and no association between exposure and preterm delivery (odds ratio = 1.0 for the highest exposure level; 95% confidence interval = 0.3-2.8). The small number of adverse outcomes reduced the precision of risk estimates, but these data indicate a potentially important relation between third trimester exposure to trihalomethanes and retarded fetal growth.