The aim of this study was to test the intervalidity of three different nutrition screening tools towards a broad population of elderly hospitalized patients. The association with risk factors and mortality was investigated. This is a prospective cohort study in three medical, surgical and geriatric settings, in Denmark and Sweden. Patients >65 years were consecutively included. Patients were screened by mini-nutritional assessment (MNA), malnutrition universal screening tool (MUST) and nutritional risk screening (NRS-2002). Anthropometrics, cognitive test (SPMSQ), as well as a questionnaire investigation regarding eating problems and life situation, were performed. Mortality within 12 months was investigated. In total, 233 patients mean (SD) age 81(7.64) years were included. A large variation in prevalence of nutritional risk was determined between the screening tools, MNA was 68% vs. MUST, 47% and NRS 54%, p < 0.0001. An overall agreement of 67% was seen (κ 0.52-0.55). Risk factors were associated with nutritional risk, including depressive mood. Only handgrip strength, fungus in mouth, serum albumin, CRP and cognitive function were associated with mortality. Fungus had the strongest association (OR 3.7; CI 1.19-11.30). The overall mortality rate was 27% during 12 months. However, none of the three screening tools predicted 12-month mortality. The findings show great variation in the prevalence of nutritional risk of under nutrition both between the tools and the settings. The level of agreement between the tools was moderate, and none of the three tools were capable of predicting 12-month mortality. A functional and psychological evaluation including oral health seems recommendable in elderly patients at nutritional risk.