Hypoxia is common in many solid tumours, including breast cancer. Hypoxia triggers the expression of hypoxia inducible factor-1alpha (HIF-1alpha), and HIF-1alpha has been associated with an impaired prognosis in breast cancer and down-regulation of the oestrogen receptor (ER), potentially affecting the treatment efficiency of antioestrogens. The role of HIF-1alpha regarding prognostic and treatment predictive information in breast cancer has not been established and we therefore analyzed HIF-1alpha using immunohistochemistry in a cohort of 377 premenopausal stage II breast cancers arranged in a tissue microarray. The patients were included in a randomized trial with either 2 years of tamoxifen or no adjuvant treatment. The tamoxifen treatment effect could be studied in subgroups of breast cancer and pure prognostic information could be scrutinized for untreated control patients. HIF-1alpha was scored as positive in 24% of the tumours and correlated positively to tumour size, Nottingham histological grade (NHG), Ki-67, Her2 and cyclin E expression and negatively to lymph node status, cyclin D1, ER and PR (progesterone receptor) expression. Surprisingly, there was no difference in tamoxifen response for patients with high or low HIF-1alpha expressing tumours. In lymph node-positive patients as well as NHG 1/2 tumours, high HIF-1alpha protein expression was significantly associated with an impaired recurrence-free survival (p=0.014, 0.018). When analyzing the subgroup of NHG 1/2 tumours, a high HIF-1alpha expression was the only independent significant prognostic marker in multivariate analysis, including standard prognostic markers, suggesting that HIF-1alpha might be a useful prognostic marker in this subgroup of breast cancer, with a rather good but diverse prognosis.