Within an ongoing multicentre phase 3 randomised trial (ELDA, cancertrials.gov ID: NCT00331097), early breast cancer patients, 65-79 years old, with average to high risk of recurrence, are randomly assigned to receive CMF (cyclophosphamide 600 mg/m2, methotrexate 40 mg/m2, fluorouracil 600 mg/m2, days 1-8) or docetaxel (35 mg/m2 days 1-8-15), every 4 weeks. Here we report an unplanned safety analysis prompted by an amendment introducing creatinine clearance as a tool to adjust methotrexate dose. Before such change, 101 patients with a median age of 70 were randomly assigned CMF (53 patients) or docetaxel (48 patients). At least one grades 3-4 toxic event of any type was reported in 40 (75.5%) and 19 (39.6%) patients with CMF and docetaxel, respectively (p=0.0002). Grades 3-4 hematological events were observed in 37 (69.8%) vs. 4 (8.3%) cases (p<0.0001) and grades 3-4 non-hematological toxicity in 12 (22.6%) vs. 15 (31.2%) patients (p=0.11), with CMF and docetaxel, respectively. A higher incidence of anemia, neutropenia, thrombocytopenia and febrile neutropenia was reported with CMF. Constipation, mucositis, nausea and vomiting were more common with CMF; diarrhoea, abdominal pain, dysgeusia, neuropathy and liver toxicity were more frequent with docetaxel. No significant interaction was found between the occurrence of severe toxicity and baseline variables, including creatinine clearance and geriatric activity scales. In conclusion, weekly docetaxel appears to be less toxic than CMF in terms of hematological toxicity.