We measured levels of cholesterol and its oxidation products, 5,6 alpha- and beta-epoxides and their common hydrolysis product cholestane triol, in breast fluids of women without breast disease, compared these levels to serum cholesterol levels, and explored associations of these breast fluid measurements with known breast cancer risk factors and other characteristics. Subjects were 105 women with no history of breast disease from whom breast fluid could be obtained by nipple aspiration. The four breast fluid measurements were significantly correlated with each other (P less than 0.0001) but none was correlated with serum cholesterol. In subsequent analyses restricted to breast fluid cholesterol and cholesterol beta-epoxide, cholesterol levels (but not beta-epoxide levels) increased with age and were higher in white than nonwhite women. Both measurements were low in women who were lactating, who were parous, or who had breast-fed. The lower levels among parous women persisted for at least 2 years postpartum or postlactation. Because breast fluid levels of cholesterol beta-epoxide are reduced for some time following a birth or cessation of lactation, the alveolar-ductal systems of parous women presumably have less cumulative exposure to this potentially carcinogenic substance. This biochemical mechanism may, in part, explain the observed reduction of breast cancer risk associated with parity.