Reporting consistency was examined in a case-control study of spontaneous abortion in Santa Clara County, California. Each case (n = 100) and two pregnant controls (n = 200), frequency-matched by last menstrual period, were interviewed twice: first after the case's spontaneous abortion (on average, 24 weeks after the last menstrual period) and again after completion of the controls' pregnancies (on average, 48 weeks after the last menstrual period). Because of concern about differential reporting of water consumption in regions with publicized water contamination, interviews included detailed questions about consumption of tap water and bottled water during pregnancy, as well as other exposures. Most factors such as caffeine consumption, cigarette smoking, employment, and pregnancy history were consistently reported between interviews and did not appear to be subject to differential reporting between cases and controls. When variables were examined by univariate analysis, controls deleted reports of tap water consumption (any vs. more) more often than did cases. There was also a suggestion of differential reporting of up to two glasses per day for tap water and bottle water consumption when they were examined as continuous variables. However, the degree of differential reporting was not sufficient to appreciably alter the measures of association between water consumption during pregnancy and spontaneous abortion.