The antiestrogen tamoxifen, a major endocrine therapy of estrogen receptor (ER)-positive breast cancer, is nevertheless inefficient in 30 to 40% of cases for unknown reasons. We retrospectively studied 50 ER-positive primary breast carcinomas. All of the patients had received tamoxifen as the only adjuvant therapy. They were divided into two groups depending on whether they relapsed within 5 years (16 tamoxifen-resistant cases) or did not relapse within 5 years (34 tamoxifen-sensitive cases). The expression of total ER beta protein, and of ER beta cx protein, was estimated anonymously in formalin-fixed, paraffin-embedded tumor sections, by using specific antibodies and quantifiying nuclear immunostaining with a computer image analyzer. All of the tumors were found to be HER-2/neu-negative by immunohistochemistry. Univariate analysis showed that Scarff-Bloom-Richardsson grade modified by Elston (SBR grade; P < 0.001), tumor size (P = 0.042), and MIB-1 proliferation index (P = 0.02) were significantly higher in tamoxifen-resistant tumors. A low level of total ER beta, whether in percentage of positive cells or in quantitative immunocytochemical (QIC) score, was also associated with tamoxifen resistance (P = 0.004). ER beta cx expression and lymph node status were similar between the two groups. The expression of ER beta in the total population was positively correlated with ER beta cx (r = 0.63, P < 0.001), and was independent of the other parameters. In a multivariate analysis, ER beta expression was the most important variable (P = 0.001), followed by SBR grade (I+II versus III; P = 0.008), and MIB-1 (P = 0.016). To conclude, tamoxifen resistance is associated with classical variables of aggressive tumors (high SBR grade, proliferation index, and tumor size) but not with node invasiveness. Low ER beta level is an additional independent marker, better than ER alpha level, to predict tamoxifen resistance.