There is evidence in the literature of a relationship between dose and response to adjuvant chemotherapy for breast cancer, although published results are conflicting. We therefore retrospectively analysed the role of dose response in patients included in four adjuvant trials of the International Breast Cancer Study Group (IBCSG, formerly the Ludwig Breast Cancer Study Group (trials I, II, III and V), all using 'classical' cyclophosphamide, methotrexate, and 5-fluorouracil (CMF). A total of 1385 node-positive patients were treated with oral cyclophosphamide, and intravenous methotrexate plus 5-fluorouracil (CMF) for at least six 4 week courses. 1350 of these were included in 6 month landmark treatment outcome analyses. A total of 1029 patients were premenopausal, 321 were postmenopausal; 800 had one to three and 550 more than three involved axillary nodes at surgery. The median follow-up ranged from 12 years for trial V to 15 years for trials I-III. Patients were grouped according to three prospectively defined dose levels based on the percentage of the protocol prescribed dose that was actually administered (level I > or = 85%, level II 65-84%, level III < 65%). Patients who received dose level II had a higher disease-free (P = 0.07) and overall survival (P = 0.03) than those who received a higher (level I) or lower (level III) percentage. The 10 year overall survival was 60% for dose level II, 56% for dose level I, 51% for dose level III. The results were generally consistent within trial, menopausal status, and oestrogen receptor status groups. The results within nodal groups showed a large difference among the dose levels for the group with one to three positive nodes (P = 0.02), but no difference for the group with four or more positive nodes. Our results indicate that the dose-response effect remains a crucial factor in adjuvant chemotherapy of breast cancer. Reductions larger than 35% in the dose administered of oral CMF adversely influenced the outcome of breast cancer patients and should be avoided. The better outcome of the intermediate dose group indicates the need to investigate other aspects involved in the cytotoxicity of adjuvant CMF chemotherapy.