Thirty-three histologically diverse bone and soft tissue tumors were analyzed biochemically for the presence of estrogen receptor protein (ERP) and progesterone receptor by means of a conventional, commercially available, steroid-binding assay (dextran-coated charcoal method) on fresh frozen tissue. These results were compared with analysis of ERP by using a specific monoclonal antibody both in an enzyme immunoassay and on frozen tissue sections by using immunohistochemical procedures. Frozen tissue sections were also examined for the presence of estrogen and progesterone receptors using fluorescein-labeled steroids. Six of the 33 tumors (18%) contained low levels of ERP ranging from 19 to 73 fmol/mg as determined by the dextran-coated charcoal method. The remaining 27 cases contained no (less than 10 fmol/mg) ERP. The ERP-positive group included a fibromatosis, leiomyosarcoma, liposarcoma (2 cases), neural sarcoma, and a synovial sarcoma. Four were high grades sarcomas, and two were low grade sarcomas. There was excellent agreement between the ERP levels determined by the dextran coated charcoal method and those determined by enzyme immunoassay. ERP could not be demonstrated immunohistochemically on frozen tissue sections of the tumors even though it could be demonstrated in breast carcinomas serving as positive controls. The failure of the immunohistochemical technique may be related to the low levels of ERP in these tumors and the difficulty of detecting antigen at threshold levels. Cytochemical localization of receptor protein employing fluoresceinated steroids did not correlate with cytosolic ERP as determined by enzyme immunoassay or the dextran coated charcoal method. Moreover, the high level of background fluorescence gave rise to a significant amount of intraobserver and interobserver variation. Although the clinical significance of ERP protein in mesenchymal tumors is still uncertain, the present findings, coupled with various clinical observations suggesting hormonal dependency of some mesenchymal tumors, indicate that investigation of a larger group of patients amenable to statistical analysis is warranted.