Estrogen receptor (ER) immunocytochemical assay (ER-ICA) was assessed in 400 human breast carcinomas. In all cases, patient's age, tumor size, histological type and Scarff-Bloom-Richardson grade, and presence or absence of axillary lymph node metastases and of vessel invasion in tumor borders were recorded. In 310 cases estrogen and progesterone receptors were concomitantly evaluated (dextran coated charcoal method). In 60 of these cases the ER immunoenzymatic assay (ER-IEA) was also assessed. Monoclonal H222sp gamma and peroxidase antiperoxidase procedures (Abbott kit) were applied in frozen sections, tumor imprints, and fine-needle aspirates. A computerized system of image analysis referred to as SAMBA (TITN), permitted a multiparametric quantitative analysis of ER-positive surfaces. With this system, in each tumor, the cellularity, percentage of ER surface versus the total cell surface and versus the epithelial (keratin-positive) surface, integrated optical density, mean optical density, index of the concentration of labeled objects, and integrated optical density histograms, were obtained and correlated to histological and biochemical data. It was shown that (a) ER antigenic sites were heterogeneously distributed in ER-positive tumors, with a specific nuclear localization in epithelial cells; (b) SAMBA 200 multiparametric analysis of the ER sites distribution in tissue was appropriate, accurate, reproducible, and therefore more reliable than the semiquantitative analysis; (c) standardization and complete automation of this method of immunoprecipitates evaluation on tissue section permits daily and routine analysis of a large number of preparations; (d) there was a correlation between ER binding sites evaluation (dextran coated charcoal) and ER antigenic sites immunodetection (ER-ICA and ER-IEA); (e) there was a correlation between the SAMBA evaluation of ER-ICA and other histological prognostic factors such as small tumor size, low Scarff-Bloom-Richardson grade; (f) the preliminary SAMBA analysis of ER-ICA in tissue sections, imprints, and fine needle aspirates suggest that fine needle aspirates may not reflect accurately the tumor cell heterogeneity.