Background: Nutrition screening is required for early identification and treatment of patients at risk for malnutrition so that clinical outcomes can be improved and health care costs reduced. Objective: To determine the criterion (concurrent and predictive) validity of the Malnutrition Screening Tool (MST) and Mini Nutritional Assessment-Short Form (MNA-SF) in older adults admitted to inpatient rehabilitation facilities. Design: Observational, prospective cohort. Participants/setting: Participants were 57 adults aged 65 years and older (mean±standard deviation age=79.1±7.3 years) from two rural rehabilitation units in New South Wales, Australia. Main outcome measurements: MST; MNA-SF; International Statistical Classification of Diseases and Health Related Problems, 10th revision, Australian Modification (ICD-10-AM) classification of malnutrition; rehospitalization; admission to a residential aged care facility (institutionalization); and discharge location. Statistical analysis performed: Measures of diagnostic accuracy with 95% CIs generated from a contingency table, Mann-Whitney U test, and χ2 test. Results: When compared with the ICD-10-AM criteria, the MST showed stronger diagnostic accuracy (sensitivity 80.8%, specificity 67.7%) than the MNA-SF (sensitivity 100%, specificity 22.6%). Neither the MST nor the MNA-SF was able to predict rehospitalization, institutionalization, or discharge location. Conclusions: The MST showed good concurrent validity and can be considered an appropriate nutrition screening tool in geriatric rehabilitation. The MNA-SF may overestimate the risk of malnutrition in this population. The predictive validity could not be established for either screening tool.