The importance of estrogen receptor (ER) determination in breast cancer is well established. Approximately 70% of ER-positive tumors are hormone responsive compared to 5-10% of ER-negative tumors. However, one-third of ER-positive tumors fail to respond, and the reasons for this are unclear. To further investigate these relationships we have determined levels of ER protein and mRNA in a number of human breast cancer biopsies. ER protein was estimated by the dextran-coated charcoal steroid binding method and by an ER immunocytochemical assay using a specific monoclonal antibody. A complimentary DNA clone (lambda OR3) encoding part of the human ER was used to determine mRNA levels. Dot blot analysis of twenty-seven tumors revealed a close agreement between ER mRNA and the dextran-coated charcoal assay (rs = 0.9; P less than 0.001). ER immunocytochemical assay staining also correlated with ER mRNA in twenty-five cases (rs = 0.75; P less than 0.001). Tumors from postmenopausal patients contained much higher levels of ER mRNA and ER protein than their premenopausal counterparts. ER-negative tumors produced no measurable ER mRNA. Northern blot analysis revealed a 6.4- and 3.7-kilobase species in ER-positive tumors and also in the human breast cancer cell line MCF-7. No differences in transcript sizes were found in tumors from hormone-responsive patients compared to nonresponding patients. We have also demonstrated, in tissue sections of normal and malignant breast, localization of ER mRNA by in situ hybridization to the same population of cells which exhibit immunoreactive ER.