Background & aims: the obesity survival paradox is an emergent issue in oncology, but its existence remains unclear particularly in older cancer patients. We aimed to assess the obesity survival paradox in older cancer patients.Methods: all consecutive cancer outpatients 65 years and older referred for geriatric assessment (GA) before a decision on cancer treatment between November 2013 and September 2016 were enrolled in the PF-EC cohort study. The main outcome was 6-month mortality. A Cox univariate and multivariate proportional hazard regression models were performed with baseline GA, oncological variables (cancer site, extension and treatment modalities) and C-reactive protein (CRP). We assessed the prognostic value of body mass index categories (i.e. malnutrition <21, 21 <= normal weight <= 24.9, 25 <= overweight <= 29.9 and obesity >= 30 kg/m(2)) in the whole study population and according to the metastatic status.Results: 433 patients with a mean age of 81.2 +/- 6.0 years were included, 51% were women, 44.3% had digestive cancers, 18% breast cancer and 14.5% lung cancer and 45% metastatic cancers. Eighty-eight of these patients (20.3%) were obese at baseline. Mortality rate was 17% during the 6-month follow-up period. After adjustment for sex, gait speed, Mini-Mental State Examination, cancer site and exclusive supportive care, obesity (compared to normal weight) was independently and negatively associated with 6-month mortality only in metastatic patients (aHR 0.17, 95% CI [0.03-0.92], P = 0.04).Conclusion: our study confirms the obesity survival paradox in older cancer patients only in the metastatic group. (C) 2018 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism.