Since 1996 when estrogen receptor beta(ER beta) was discovered, much effort has been devoted to the question of the value of ER beta as a prognostic and/or predictive factor in breast cancer and its potential as a novel target for pharmacological intervention. When estrogen receptors are applied on sucrose gradients and quantified by ligand binding, we found that in contrast to ER alpha, which has a narrow tissue distribution, ER beta is expressed in many tissues including both normal and malignant breast tissue. Receptor protein levels in tissues can also be measured from the intensities of bands after Western blotting and can be quantified when purified and quantified receptor is used as a standard. With this technique, we found that there were some tumors which had over 600 fmol/mg of ER beta protein but no detectable estradiol binding. In such tumors, RT-PCR analysis revealed that ER beta cx is the only ER beta isoform present. ER beta cx is a splice variant which utilizes an alternative exon 8. This change in the C-terminus results in very poor binding to estradiol (E2) and has a dominant negative effect on ER alpha function. Immunohistochemical analysis with an ER beta cx specific antibody in 115 ER alpha-positive breast cancers revealed that about half of the samples expressed ER beta cx protein. Initial analysis of samples from patients with preoperative tamoxifen treatment revealed that ER alpha-positive tumors expressing ER beta cx and lacking PR seemed to be resistant to the anti-estrogen. We conclude that, in order to better characterize breast cancers and design appropriate therapy for individual patients, assays for ER beta cx must be made available to clinicians.