Purpose Age induces a progressive decline in functional reserve and impacts cancer treatments. Telomere attrition leads to tissue senescence. We tested the hypothesis that telomere length (TL) could predict patient vulnerability and outcome with cancer treatment. Patients and methods An ancillary study in the Elderly Women GINECO Trial 3 was performed to evaluate the impact of geriatric covariates on survival in elderly advanced ovarian cancer patients receiving six cycles of carboplatin. TL was estimated from peripheral blood at inclusion using standard procedures. Results TL (in base pairs) was estimated for 109/111 patients (median 6.1 kb; range [4.5-8.3 kb]). With a cut-off of 5.77 kb, TL discriminated two patient groups, long telomere (LT) and short telomeres (ST), with significantly different treatment completion rates of 0.80 (95%CI [0.71-0.89]) and 0.59 (95%CI [0.41-0.76]), respectively (odds ratio [OR]=2.8, p=0.02). ST patients were at higher risk of serious adverse events (SAE, OR=2.7; p=0.02) and had more unplanned hospital admissions (OR=2.1; p=0.08). After adjustment on FIGO stage, TL shorter than 6 kb was a risk factor of premature death (HR=1.57; p=0.06). Conclusion This exploratory study identifies TL as predictive factor of decreased treatment completion, SAE risk, unplanned hospital admissions and OS after adjustment on FIGO stage.