Chlorination by-products (CBPs) in drinking water have been associated with an increased risk of adverse pregnancy outcomes, including small for gestational age at term (term-SGA) and preterm delivery. Epidemiological evidence is weakened by a generally inaccurate exposure assessment, often at an ecological level. A case control study with incident cases was performed in nine Italian towns between October 1999 and September 2000. A total of 1,194 subjects were enrolled: 343 preterm births (26th-37th not completed week of pregnancy), 239 term-SGA (from 37th completed week, and weight less than the lowest 10th percentile) and 612 controls. Exposure was assessed both by applying a questionnaire on mothers' personal habits during pregnancy and by water sampling directly at mothers' homes. Levels of trihalomethanes (THMs) were low (median: 1.10 microg l(-1)), while chlorite and chlorate concentrations were relatively high (median: 216.5 microg l(-1) for chlorites and 76.5 microg l(-1) for chlorates). Preterm birth showed no association with CBPs, while term-SGA, when chlorite levels > or =200 microg l(-1) combined with low and high levels of inhalation exposure are considered, suggested a dose-response relationship (adjusted-Odds Ratios (ORs): 1.52, 95%CI: 0.91-2.54 and 1.70, 95%CI: 0.97-3.0, respectively). A weak association with high exposure levels of either THMs (> or =30 microg l(-1)), or chlorite or chlorate (> or =200 microg l(-1)) was also found (adjusted-OR: 1.38, 95%CI: 0.92-2.07). Chlorine dioxide treatment is widespread in Italy; therefore, chlorite levels should be regularly and carefully monitored and their potential effects on pregnancy further evaluated and better understood.