Benefit from chemotherapy treatment in breast cancer patients is determined by the molecular make-up of the tumour. In a retrospective analysis, we determined the molecular subtypes of breast cancer originally defined by expression microarrays by immunohistochemistry in tumours of patients who took part in a randomised study of adjuvant high-dose chemotherapy in breast cancer. In addition, the topoisomerase II alpha (TOP2A) amplification status was determined by fluorescence in situ hybridisation and chromogenic in situ hybridisation. 411 of the 753 tumours (55%) were classified as luminal-like, 137 (18%) as basal-like and 205 (27%) as human epithelial receptor type 2 (HER2) amplified. The basal-like tumours were defined as having no expression of ER and HER2; 98 of them did express epidermal growth factor receptor and/or cytokeratin 5/6. The luminal-like tumours had a significantly better recurrence free and overall survival than the other two groups. From the 194 HER2-positive tumours, 47 (24%) were shown to harbour an amplification of TOP2A. Patients with an HER2-amplified tumour randomised to the high-dose therapy arm did worse than those in the conventional treatment arm, possibly caused by the lower cumulative anthracycline dose in the high-dose arm. The tumours with a TOP2A amplification contributed hardly to this difference, suggesting that TOP2A amplification is not the cause of the steep dose-response curve for anthracyclines in breast cancer. Possibly, the difference of the cumulative dose of only 25% between the treatment arms was insufficient to yield a survival difference.