Estrogen receptor (ER) analysis by sucrose density gradient analysis (SDGA) was compared with histochemical localization of estrogen binding using 6-carbomethoxy-bovine serum albumin-fluorescein isothiocyanode-estradiol (E2-6-CMO-BSA-FITC) (6FE), 17-thiosemicarbozene-BSA-FITC-estrogen (E-17-TSC-BSA-FITC) (17FE), and polyestradiol phosphate-antiestradiol antibody-FITC (PEP) on serial frozen sections of metastatic breast cancer lesions from 72 patients treated with hormonal therapy. A comparison of assays to clinical responses gave the following sensitivities: ER-SDGA, 90%; 6FE, 50%; 17FE, 55%; and PEP, 58%. Specificities were as follows: ER-SDGA, 81%; 6FE, 47%; 17FE, 41%; and PEP, 59%. The SDGA gave the highest predictive value for clinical response, while the predictive value for each of the three histologic techniques approximated that predicted by chance alone. These selected histochemical techniques for ER localization, despite impressive cytologic localization patterns, therefore, do not correlate with clinical response.