BACKGROUND AND AIMS: Malnutrition is related to increased mortality and poor outcome. Therefore, early diagnosis and intervention of malnutrition are recommended. However, the optimal nutrition screening tool for diagnosing malnutrition remains undetermined. This study aimed to verify the discrimination and difficulty of nutrition screening items through item response theory (IRT) analysis and develop a simpler malnutrition screening tool. METHODS: This study enrolled 10,375 patients aged ≥18 years who were admitted at Tokushima University Hospital. Patients who were pregnant had short-term hospitalization (≤3 days), were hospitalized only in the weekend, could not hear clearly, and were hospitalized merely for examination were excluded. A skilled dietitian performed the Subjective Global Assessment, rating a good nutritional status as A, moderate malnutrition as B, and severe malnutrition as C. RESULTS: According to Subjective Global Assessment, we classified 7119 patients as A, 2892 as B, and 364 as C. Between these classes, the total number of application items was significantly lower in class A but significantly higher in class C. In the discrimination of each item calculated by IRT analysis, the highest discrimination item was "Has your food intake been less than usual?", followed by "Have you had anorexia?", "Have you had vomiting?", "Have you had nausea?", "Have you had dehydration?", and "Have you lost weight?". CONCLUSIONS: Human Key Tool of Nutrition, which is based on the IRT analysis results, is a new simple nutrition screening tool developed by us. Human Key Tool of Nutrition can contribute to the widespread use of nutrition screening.