The aim of the study was to achieve the external validation of the SAGA instrument on the SAFES cohort. This was a prospective longitudinal multicentre cohort study, including patients aged 75 years or over, hospitalized in a short stay medical ward via emergency department. A comprehensive geriatric assessment was implemented. The psychometric validation enabled the study of feasibility, internal consistency, convergent, discriminant, and predictive validity of the instrument. The 1 306 patients of the cohort was 85±6 years, with a majority of women (65%). The completion rate of the SEGA instrument was 94%. Internal consistency was good (Cronbach alpha coefficient=0.7). Convergent validity was poor: Donini instrument (kappa=0.18; IC 95%=0.13-0.23), Rockwood instrument (kappa=0.04; IC 95%=0.02-0.06), and Winograd instrument (kappa=0.04; IC 95%=0.01-0.07). The ability of the instrument to discriminate clinically different groups was good. The SEGA instrument predicted well one-year mortality as well as one-year institution admission. Despite poor convergent validity (which is classic with frailty tools, SEGA instrument has satisfactory metrological properties, allowing its use in emergency departments and immediate post-emergency circumstances.