The purpose of this study is to investigate EZH2 in a large series of breast cancer patients for its prognostic and predictive value, and to evaluate its functional role in treatment response in vitro. EZH2 levels were measured using quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) in primary breast cancer specimens and related to clinicopathologic factors and disease outcome. EZH2 expression was downregulated with siRNAs in MCF7, to assess expression alterations of putative EZH2 downstream genes and to determine cell numbers after treatment with the anti-estrogen ICI 164384. In 688 lymph node-negative patients who did not receive adjuvant systemic therapy, EZH2 was not significantly correlated with metastasis-free survival (MFS). In 278 patients with advanced disease treated with first-line tamoxifen monotherapy, the tertile with highest EZH2 levels was associated with the lowest clinical benefit (OR = 0.48; P = 0.02) and with a shorter progression-free survival (PFS) in both univariate (HR = 1.80; P < 0.001) and multivariate analysis, including traditional factors (HR = 1.61; P = 0.004). In vitro, EZH2 silencing in MCF7 caused a 38% decrease in cell numbers (P < 0.001) whereas ICI 164384 treatment resulted in a 25% decrease (P < 0.001) compared to controls. Combining EZH2 silencing with ICI treatment reduced cell numbers with 67% (P < 0.001) compared to control conditions. EZH2 downregulation was associated with an almost two-fold upregulation of the estrogen receptor alpha (ER) (P = 0.001). In conclusion, EZH2 has no prognostic value in breast cancer. High levels of EZH2 are associated with poor outcome to tamoxifen therapy in advanced breast cancer. Downregulated EZH2 leads to upregulation of the ER and better response to anti-estrogens.