We recently studied pregnancies occurring during 1980-1985 in four study areas in Santa Clara County, California. Two of the areas were exposed to solvent-contaminated drinking water during 1980 and 1981, and two were unexposed. There was an overall excess of spontaneous abortions among women who reported any tapwater consumption during the first trimester of pregnancy compared with those who reported no tapwater consumption [odds ratio (OR) = 4.0; 95% confidence interval (CI) = 1.8-9.1)], regardless of exposure to the contaminated water. The odds ratio for spontaneous abortion for women reporting any vs no tapwater was 6.9 (95% CI = 2.7-17.7) after adjustment for numerous potential confounders using multiple logistic regression analyses. The elevated odds ratio of spontaneous abortion was seen among tapwater drinkers who used no filters or softener-type filters but not among women who reported use of active filters. Spontaneous abortion rates were reduced in women who reported any vs no bottled water consumption (OR = 0.26; 95% CI = 0.16-0.43). Among women who reported no tapwater consumption, no birth defects occurred among 263 live births; in comparison, among women who reported tapwater consumption, 4% of 908 live births had defects (P = 0.0001). We observed no relation between birth defects and bottled water use.