We wanted to study cyclin A as a marker for prognosis and chemotherapy response. A total of 283 women with metastatic breast cancer were initially enrolled in a randomised multicentre trial comparing docetaxel to sequential methotrexate-fluorouracil (MF) in advanced breast cancer after anthracycline failure. Paraffin-embedded blocks of the primary tumour were available for 96 patients (34%). The proportion of cells expressing cyclin A was determined by immunohistochemistry using a mouse monoclonal antibody to human cyclin A. Response evaluation was performed according to WHO recommendations. The median cyclin A positivity of tumour cells was 14.5% (range 1.2-45.0). Cyclin A correlated statistically significantly to all other tested proliferation markers (mitotic count, histological grade and Ki-67). A high cyclin A correlated significantly to a shorter time to first relapse, risk ratio (RR) 1.94 (95% CI 1.24-3.03) and survival from diagnosis, RR 2.49 (95% CI 1.45-4.29), cutoff point for high/low proliferation group 10.5%. Cyclin A did not correlate to chemotherapy response or survival after anthracycline, docetaxel or MF therapy. Of all tumour biological factors tested (mitotic count, histological grade and Ki-67), cyclin A seemed to have the strongest prognostic value. Cyclin A is a good marker for tumour proliferation and prognosis in breast cancer. In the present study, cyclin A did not predict chemotherapy response.