Although the androgen receptor (AR) is frequently expressed in breast cancer, its relevance in the disease is not fully understood. In addition, the relevance of AR in determining tamoxifen treatment efficiency requires evaluation. To investigate the tamoxifen predictive relevance of the AR protein expression in breast cancer. Patients were randomised to tamoxifen 40 mg daily for 2 or 5 years or to no endocrine treatment. Mean follow-up was 15 years. Hazard ratios were calculated with recurrence-free survival as end point. In patients with oestrogen receptor (ER)-negative tumours, expression of AR predicted decreased recurrence rate with tamoxifen (hazard ratio (HR)=0.34; 95% confidence interval (CI)=0.14-0.81; P=0.015), whereas the opposite was seen in the AR- group (HR=2.92; 95% CI=1.16-7.31; P=0.022). Interaction test was significant P<0.001. Patients with triple-negative and AR+ tumours benefitted from tamoxifen treatment (HR=0.12; 95% CI=0.014-0.95 P=0.044), whereas patients with AR- tumours had worse outcome when treated with tamoxifen (HR=3.98; 95% CI=1.32-12.03; P=0.014). Interaction test was significant P=0.003. Patients with ER+ tumours showed benefit from tamoxifen treatment regardless of AR expression. AR can predict tamoxifen treatment benefit in patients with ER- tumours and triple-negative breast cancer.