Disinfection by-products in tap water have been found in some studies to be associated with adverse pregnancy outcomes, but little is known about how water use and consumption might change during early pregnancy. Estimating water-related activities only at one time during pregnancy could easily lead to exposure misclassification. To evaluate changes in water use among pregnant women, we used data from a large epidemiologic study in which 1990 women were interviewed around 9 and 20 weeks' gestation. The water variables that were examined included ingestion of cold and hot tap water as well as of bottled water, showering and bathing. Changes were detected between early and mid-pregnancy for ingested cold tap water and showering. Thirty-three percent of the subjects changed cold-water ingestion by > or =1.0 liters/day and 44% changed their time showering by > or =35 min per week during this period. Increases in cold tap water intake were associated with age >35 years, income < $40,000, and non-Hispanic white ethnicity. We also found that the proportion of the total variation due to within-subject variability was 62% for hot tap water ingestion but only 35% for showering and approximately 50% for cold tap water, bottled water and bathing. Limited resources in epidemiologic studies often require a decision between collecting data for a large number of people or collecting multiple measurements for a smaller number of people. The results in this study will be useful to researchers who need to determine where to invest their effort when assessing water-related exposures and should help in evaluation of previously performed studies.