Environmental tobacco smoke (ETS) exposure has been studied in relation to many diseases. The ability of a study to find an association between exposure and disease is, in part, determined by the accuracy of the exposure measure. This study examined how accurately one question, the number of smokers in the household, asked at birth, predicts ETS exposure in pregnant nonsmokers as assessed by serum cotinine. Blood specimens, drawn at 15-19 weeks gestation, from 783 women who participated in a prenatal screening program in California in 1992 were analyzed for cotinine. Serum cotinine was significantly correlated with the number of smokers in the household (r=0.35, P<0.001, geometric mean cotinine (nanograms per milliliter) for 0 smokers=0.06, 1 smoker=0.18, 2 or more smokers=0.29). Using multiple regression, the number of smokers in the household accounted for 11% of the variation in serum cotinine. Cotinine concentrations were twice as high in women living with one or more smokers compared to women not living with a smoker, when reported exposure (0 or >0 h) at home, work and other places was similar. Thus, the number of household smokers can account for a statistically significant amount of variation in serum cotinine and omission of this information would result in an underestimation of ETS exposure. Although use of this question alone does not provide an adequate estimation of ETS exposure as determined by serum cotinine, the results of this study indicate that this question is an important component of assessing ETS exposure.