The authors used the Medline data retrieval system for 1966-2001 to identify and synthesize the epidemiological evidence for adverse effects of prenatal exposure to chlorination byproducts on birth defects. For 5 studies that provided estimates of the effect of exposure on 1 or several birth defects, the authors calculated summary odds ratios and 95% confidence intervals using the Mantel-Haenszel method for a fixed-effects model, and the DerSimonian-Laird method for a random-effects model. The results of this meta-analysis provide evidence for an effect of exposure to chlorination byproducts on the risk of any birth defect, and of neural tube and urinary system defects. The results for respiratory system, major cardiac, and oral cleft defects were heterogeneous and inconclusive. Differences in the susceptibility of populations, and different approaches in exposure assessment, may explain the heterogeneity of the results.