PURPOSE: A number of prospective studies have shown that adjuvant CEF significantly improves disease-free and overall survival as compared to CMF in breast cancer patients. Our aim was to determine whether the benefit of epirubicin versus methotrexate differs according to TIMP-1 tumour cell immunoreactivity. EXPERIMENTAL DESIGN: Tissue micro arrays from 647 patients randomly assigned to CMF or CEF in DBCG trial 89D were included. The primary end-point was invasive disease-free survival (IDFS). A central assessment of tissue inhibitor of metalloproteinases 1 (TIMP-1) status was performed using immunohistochemistry (IHC). Tumours were regarded as TIMP-1 positive if epithelial breast cancer cells were stained using the anti-TIMP-1 monoclonal antibody VT7. RESULTS: By central assessment 75% of tumours were classified as tumour cell TIMP-1 positive. Among CEF-treated patients, individuals with TIMP-1 negative tumours had a significant longer IDFS than patients with TIMP-1 positive tumours (p=0.047). The multivariate Cox regression analysis of IDFS showed that CEF was superior to CMF among patients with TIMP-1 negative tumours (hazard ratio (HR)=0.51; 95% confidence interval (CI): 0.31-0.84, p=0.0085), while no significant difference could be demonstrated among patients with TIMP-1 positive tumours (HR=0.88; 95% CI: 0.68-1.13, p=0.32). A non-significant TIMP-1 status (positive or negative) versus treatment (CMF or CEF) interaction was detected for IDFS (p=0.06) and OS (p=0.21). CONCLUSION: Lack of TIMP-1 tumour cell immunoreactivity seems to predict a favourable effect of epirubicin-containing adjuvant therapy in primary breast cancer. However, an independent study is awaited to validate the potential predictive value of TIMP-1 immunoreactivity