The high prevalence of malnutrition and associated adverse outcomes in older people is well documented. Early identification of malnutrition and intervention in hospital patients may improve clinical outcome. ‘Malnutrition Universal Screening Tool’ (‘MUST’) is the preferred screening method for malnutrition in UK. The Short Form Mini Nutritional Assessment (MNA-SF) has been developed specifically for older populations. Little information is available regarding the comparability of these commonly used screening instruments. The aim of this study was to compare these nutrition screening tools and evaluate their predictive validity.A prospective cross sectional study of nutritional status in 149 inpatients aged 65–99 years was carried out. Exclusion criteria were terminal illness, active malignancy, and lack of capacity to consent. Main nutritional measures were weight, height, body mass index, mid arm circumference, hand grip strength, and serum albumin levels. Data on mortality, length of stay (LOS) and hospital readmissions were collected retrospectively and related to ‘MUST’ and MNA-SF scores.The main findings were that there was a ‘moderate’ agreement between ‘MUST’ and MNA-SF [κ = 0.50, 95% CI (0.39, 0.60)], that both ‘MUST’ and MNA-SF scores predicted mortality (p = 0.013 and 0.009 respectively), and that LOS increased progressively with MNA-SF category.The MNA-SF categorises many more older people admitted to hospital as at risk of malnutrition than the ‘MUST’. Both tools have predictive validity with regard to mortality but MNA-SF better predicts length of stay and readmission rates. These findings support screening all older hospital patients for malnutrition, with either tool.