It has been shown that the risk of breast cancer developing in certain morphologically identifiable benign breast lesions correlates with expression of estrogen receptor alpha (ER-alpha). Although ER-alpha and ER-beta genes share a large degree of homology, it is generally thought that their distribution and functions are substantially different in many tissues. Recent development of reliable antibodies to ER-beta has provided this first opportunity to test the hypothesis that the likelihood of malignant transformation in morphologically benign breast lesions can be accurately defined by the distribution and level of ER-beta expression relative to that of ER-alpha. Using a monoclonal antibody, ER-beta protein expression has been analyzed in 53 normal breasts and compared with a cohort of histologically distinct breast lesions of different prognostic risk (54 hyperplasia of usual type, 35 ductal carcinoma in situ, and 141 invasive cancers). All of these tissues were also assessed for ER-alpha. Expression of ER-beta protein was also analyzed in an additional spectrum of benign breast lesions with low or negligible risk of progression to malignancy. The median proportion of cells expressing ER-beta was highest in normal breast lobules (median 94.33%, interquartile range 78.25-99.00) but declined significantly through usual ductal hyperplasia (median 76.67, interquartile range 49.17-95.00, P = 0.002) and ductal carcinoma in situ (median 70.00, interquartile range 59.00-85.00, P = 0.009) to invasive cancer (median 60.00, interquartile range 50.00-80.00, P < 0.001). An appreciable proportion (33.81%) of ER-alpha-negative invasive cancers expressed ER-beta. A high but variable level of ER-beta expression occurred in the benign lesions. The data from the intact histologic tissues were evaluated with respect to the relative expression of ER-alpha and ER-beta in five mammary cell lines of different behavioral phenotype (MCF7, ZR-75, T47D, MDAMB231, HUMA121). The highly significant differences in expression and distinct tissue distributions of ER-alpha and ER-beta within the histologic lesions of defined risk, together with the data from the cell lines, support the original hypothesis that the tissue concentration, relative occurrence, and/or interaction of these two types of estrogen receptor may play an important role in modulating mammary tumorigenesis.