BACKGROUND: Inconsistent doses and schemes are commonly used in older patients receiving cancer chemotherapy. We performed this study in patients with cancer and age ≥ 70 years to determine the frequency of undertreatment and overtreatment as well as factors influencing the decision to modify chemotherapy doses. PATIENTS AND METHODS: Patients aged ≥70 years starting new chemotherapy regimens were prospectively included in a multicentre study. The schedule and drug doses were determined by the treating oncologist. Pre-chemotherapy assessment included sociodemographics, treatment details and geriatric assessment (GA) variables. Association between these factors and undertreatment (use of less intensive cancer treatment [LICT] in a fit patient) or overtreatment (use of standard cancer treatment in an unfit older patient) were examined by multivariate logistic regression. RESULTS: Three- hundred ninety-seven patients were included, 43% of whom received LICT. If not adjusted for GA, toxicity did not differ between those receiving LICT (38%) or standard doses of chemotherapy (37%). If the dose of chemotherapy was analyzed according to the results of GA 61 (15%) patients had been undertreated and 133 (34%) had been overtreated. Undertreatment was related with increasing age and decreased renal function. Factors related with overtreatment were younger age, curative intention of treatment, prescription of G-CSF as primary prophylaxis and adequate cognitive status. Overtreated patients had more grade 3-4 toxicity than those receiving treatment adapted to fragility (42% vs 31%; p < 0.05). CONCLUSIONS: The use of chemotherapy without considering GA leads to overtreatment more commonly than undertreatment in older patients with cancer. Oncologists should take into account the results of GA to stratify patients and to avoid under or overtreatment.