The clinical response of advanced endometrial adenocarcinoma to progestin therapy does not correlate perfectly with biochemically assayed progesterone receptor status of the tumor. We have previously suggested that heterogeneity of progesterone receptor at the cellular, tumor, and tissue levels, not detectable by the biochemical assay, might contribute to this discrepancy. A monoclonal antibody, hPRa-1, generated against human progesterone receptor, was used in the present study to immunohistologically define the heterogeneity of progesterone receptor distribution in primary endometrial carcinomas. Twenty-four hysterectomy specimens removed for the treatment of endometrial adenocarcinoma were examined by biochemical assay of progesterone receptor and immunohistochemistry. In two cases, in which tumor occupied almost all of the endometrial lining, more extensive sampling was performed with removal of four noncontiguous sites. Each site was subdivided for immunohistochemistry and biochemical assay of progesterone receptor. When present, progesterone receptor localization was confined to the nuclei of target cells. Variability in the distribution and intensity of staining was consistently observed within the tumors. Of 24 tumors 15 were determined to be progesterone receptor positive by biochemical assay, while 12 of 24 tumors displayed immunolocalization for progesterone receptor. The correlation of the results by the two methods was high (20 of 24 cases, 83%), and the discrepancies in three cases appeared to reflect tissue and tumor heterogeneity. Immunolocalization has demonstrated that heterogeneity is present at the tumor, tissue, and cellular level within endometrial carcinomas, and the failure of some progesterone receptor-positive (by biochemical assay) tumors to respond to progestin therapy may reflect false positive results due to contamination of progesterone receptor-negative tumors by adjacent benign endometrium or myometrium.