Analysis of estrogen receptor protein (ERP) was carried out on 784 patients with primary breast carcinomas. Significant relationships were found with the following epidemiologic features; age at diagnosis, menstrual status, race, and use of exogenous hormones (contraceptive and menopausal estrogens). Patients taking either type of hormone at diagnosis had lower median total ERP binding levels. Although the data do not explain how recent hormone usage affects tumor ERP, it seems advisable to discontinue these hormones as early as possible before a breast biopsy because they may cause spuriously low ERP levels. Correlation with pathology variables confirmed most prior observations relating to tumor type, histologic grade, and lymphocytic infiltrate. However, in this large series, infiltrating lobular carcinomas were not ERP-positive more frequently than duct carcinomas to any significant degree. The pattern of progesterone receptors (PRP) did not vary significantly with menstrual status, but in other respects correlations of epidemiologic and pathologic variables were similar for ERP and PRP.