Objective To investigate which one out of the 2 most commonly used tools for screening malnutrition, Mini Nutritional Assessment (MNA, specially designed for the elderly) and nutritional risk screening 2002 (NRS 2002, usually for inpatients) is more suitable for the geriatric inpatients. Methods A total of 179 inpatients over 65 years old from geriatric ward in our hospital from July 2010 to January 2012 were enrolled in this study. All the patients were assessed with MNA-short form (SF) and NRS2002 within 48h after admission. The coincidence of the 2 tools was studied, and their efficiency of guiding the patients for nutritional support was compared. Meanwhile, the relationship of the nutritional status screened by the 2 methods with traditionally used nutritional indicators, i.e. body mass index (BMI), serum albumin and pre-albumin was analyzed. Results By MNA-SF, 55 cases (30.7%) were rated as at risk of malnutrition, and 42 cases (23.5%) were as malnutrition, which were significantly different from the results by NRS2002 (69 cases as at risk of nutrition, 38.5%). MNA-SF had an efficiency of 29.1% for the patients at risk of malnutrition, and of 50% for those of malnutrition, and NRS 2002 had an efficiency of 49.3%. Besides, the commonly used nutritional indicators BMI, serum albumin and pre-albumin showed little correlation with the results rated by MNA-SF and NRS2002 (r＝0.09 to 0.48), indicating that these indicators had no reference value for nutritional risks screening. Conclusion In the nutritional risks screening for geriatric inpatients in department of internal medicine, MNA-SF is a relatively sensitive tool for screening nutritional risks, while, NRS2002 has a better value in the guidance for the patients needing for nutritional therapy. The conventionally used nutritional indicators such as BMI, serum albumin and pre-albumin are not reliable or effective factors for screening nutritional risks in the elderly.