Prior reports identify higher serum concentrations of estrogens and androgens as risk factors for breast cancer, but steroids in nipple aspirate fluid (NAF) may be more related to risk. Incident breast cancer cases and mammography controls were recruited. Sex steroids were measured in NAF from the unaffected breasts of cases and one breast of controls. Menopausal status and menstrual cycle phase were determined. NAF steroids were purified by HPLC and quantified by immunoassays. Conditional logistic regression models were used to examine associations between NAF hormones and case-control status. NAF samples from 160 cases and 157 controls were evaluable for hormones. Except for progesterone and dehydroepiandrosterone (DHEA), the NAF and serum concentrations were not significantly correlated. NAF estradiol and estrone were not different between cases and controls. Higher NAF (but not serum) DHEA concentrations were associated with cases, particularly among estrogen receptor (ER)-positive cases (NAF odds ratio (OR) = 1.18, 95 % confidence interval (CI) 1.02, 1.36). NAF DHEA was highly correlated with NAF estradiol and estrone but not with androstenedione or testosterone. Higher progesterone concentrations in both NAF and serum were associated with a lower risk of ER-negative cancer (NAF OR = 0.69, 95 % CI 0.51, 0.92). However, this finding may be explained by case-control imbalance in the number of luteal phase subjects (2 cases and 19 controls). The significantly higher concentration of DHEA in NAF of cases and its correlation with NAF estradiol indicates a potentially important role of this steroid in breast cancer risk; however, the negative association of progesterone with risk is tentative.