Reproductive breast cancer risk factors are hypothesized to act by increasing exposure of the breast to endogenous estrogens, but few studies have quantitatively examined the association of these risk factors with breast tissue composition. This study is part of a case-control study of breast histological characteristics and breast cancer risk, nested within the Nurses' Health Study, a prospective study of 121,700 registered nurses. We studied 300 women who had not been diagnosed with breast cancer, but for whom we obtained slides from a prior benign breast biopsy. We used a computer-assisted image analysis technique to assess the proportion of epithelial and fibrous stromal tissue on benign breast biopsy slides, excluding obvious mass lesions. Mean epithelial proportion was 5.3% (0.1-23%), and mean stromal proportion was 58.7% (3-93%). Women with proliferative breast disease without atypia had higher epithelial and stromal proportions than women with nonproliferative breast disease (P < 0.001). Postmenopausal women had a lower epithelial proportion (P = 0.01), and increasing age at biopsy was associated with decreasing stromal proportion among postmenopausal parous women (P = 0.004). Among premenopausal women, increasing years since last birth was associated with lower epithelial proportion (P < 0.001). Other reproductive risk factors were not independently associated with epithelial or stromal proportion. Epithelial and stromal breast tissue were associated with different factors with the exception of proliferative breast disease, which was associated with an increase in both epithelial and stromal proportion. The quantitative measurement of epithelial and stromal proportion may be useful for measuring changes in breast composition.