We conducted a population-based case control study of neural tube defects and drinking water contaminants, specifically, disinfection by-products. We used public monitoring records concurrent with the first month of gestation to assess exposure. The prevalence odds ratios (PORs) for the highest tertile of total trihalomethanes compared with the lowest was 1.6 (95% confidence interval [CI] = 0.9-2.70). Surface water source was also associated with neural tube defects (POR = 1.5; 95% CI = 0.9-2.5). Sensitivity analyses restricted to isolated neural tube defect cases and mothers with known residence at conception yielded stronger associations [total trihalomethanes, POR = 2.1 (95% CI = 1.1-4.0); surface water, POR = 1.7 (95% CI = 0.9-3.2)]. Other major groups of disinfection by-products (haloacetic acids and haloacetonitriles) showed little relation to these defects.