During 1980-1981, solvents leaked from an underground storage tank of a semiconductor firm in southern Santa Clara County, California, contaminating local drinking water. The contaminated well was closed in December 1981. An epidemiologic study conducted in 1983 confirmed statistically significant excesses of adverse pregnancy outcomes in an exposed community compared with an unexposed community, but could not establish a causal connection between the leak and the adverse outcomes. This study expanded the first study; adverse pregnancy outcomes occurring in 1980-1985 were studied in two communities exposed to the contaminated drinking water and in two demographically comparable but unexposed communities. The period 1980-1981 was the time period in which the well was considered to have been contaminated and 1982-1985 was considered the postcontamination time period. Both exposed and unexposed communities were considered unexposed during the latter period (1982-1985). Out of 10,055 households surveyed, interviews were conducted with 1,105 women who reported one or more eligible pregnancies. Miscarriages and birth defects were validated by medical record review or physician reports. Although the authors again observed statistically significant excesses of spontaneous abortions and birth defects in the originally studied exposed area in 1980-1981, they observed deficits of these outcomes in the second exposed study area. Adjustment for potential confounders did not alter these findings. Analyses of pregnancy outcomes during 1981 in relation to exposure estimates based on hydrogeologic modeling of water and contaminant distribution within the exposed areas also indicated that the leak was not likely to have caused the observed excesses of adverse pregnancy outcomes in the originally studied area.