As part of our work on the influence of water source on reproductive outcome, Sprague-Dawley rats were randomized to tap water, bottled water, or deionized water treatment groups, utilizing 160 animals per treatment; animals received the water prior to and during pregnancy. Rats were shipped in four batches (A-D). Batch effects were seen for several reproductive parameters. Because the tap water supply was interrupted by an earthquake resulting in an unbalanced design, primary analyses utilized only batches C and D, which included most of the tap water-treated rats. A treatment effect with respect to resorption frequency was seen that was marginally significant using a fixed-effects analysis of variance (P = 0.053), but not when batch was entered as a random effect (P = 0.36). The data were modeled by logistic regression, controlling for batch, litter size, and batch-treatment interaction. The odds ratio comparing tap to bottled water was 1.8 (95% CI 1.0 to 3.3, P = 0.05), which was similar to the epidemiologic result that prompted this study. The magnitude of this association varied by batch, and the difference in resorption frequency was within the range of variation seen for control animals. Although these findings do not justify public health action at this time, further investigation is warranted.