The relationship of passive smoking to diet was examined in 82 female nonsmokers who provided a quantitative diet history in 1986. Exposure to environmental tobacco smoke (ETS) was assessed by urinary cotinine measurement. Mean values for each dietary variable, adjusted for age, ethnicity, education, and last week's ethanol intake, were compared among unexposed women and women with low or high ETS exposure. Linear relationships with amount of ETS exposure were also sought. Intakes of beta-carotene and cholesterol were found to be inversely related to ETS exposure. Since these nutrients have been associated with lung cancer risk, they are potential confounders of the passive-smoking/lung-cancer relationship. Although we estimate the confounding effect of these dietary factors to be modest, they should be measured carefully in future studies of this relationship.