BACKGROUND: Adjuvant chemotherapy in elderly women is currently perceived as one of the priorities in breast cancer (BC) research and, to date, we lack practical guidelines in this age group. Therefore we performed a retrospective analysis of the actual use of adjuvant chemotherapy according to each negative prognostic factor. PATIENTS AND METHODS: Charts of all consecutive elderly patients aged 70 years or more with operable BC referred to our institution between 1999 and 2003 were reviewed for tumour stage and treatment, and compared with an equal cohort of younger randomly selected postmenopausal patients (control group). RESULTS: A total of 260 elderly patients (mean age 75.6 years, age range 70-97 years) with histological diagnosis of early BC were eligible. Conserving surgery was performed in 54.6% of patients, nodal dissection in 84.6% and sentinel node biopsy in 5.8%. Tumour size was pT2-pT3 in 45.4% of patients; grading was G3 in 27.3%, hormonal status was negative (HR-) in 16.9% and lymph nodes were involved N+ in 36.1%. Of 188 patients presenting one or more risk factors (pT2-3, G3, N+, HR-), 48.4% were not proposed for adjuvant chemotherapy (compared with 7.2% in the control group), 39.8% of those with nodal involvement (compared with 4.3% of controls, P <0.0001) and 22.7% of those who were HR- (compared with 0.0% of controls, P=0.0002). Considering only patients receiving non-anthracycline-based chemotherapy, 20 elderly patients (25.9%) were unable to complete the planned number of cycles (compared with 4.7% of controls, P=0.0002). The 2-year disease-free survival was significantly decreased in N+ HR- patients compared with the remaining elderly patients (49.9% compared with 90.9%, P=0.0006). CONCLUSIONS: Elderly BC patients receive much less adjuvant chemotherapy, according to each prognostic factor. N+ HR- disease probably represents the most reasonable indication. As the toxicity of the CMF regimen frequently caused interruption of treatment, alternative regimens should be assessed in this age class.