We evaluated the Mini-Nutritional Assessment (MNA) test and the short-form MNA as screening tools for malnutrition in the Japanese elderly population.A cross-sectional study of 226 elderly Japanese patients (78.6 +/- 0.5 y of age, mean +/- standard deviation; 67 men and 159 women) in various settings was carried out. Nutritional assessment included MNA, anthropometric measurements, and biochemical markers.According to the original cutoff point of the full MNA, 19.9% of those assessed were malnourished, 58.0% were at risk of malnutrition, and 22.1% were well nourished. Significant correlations were found between full MNA scores and age (r = -0.14), body mass index (r = 0.59), serum albumin (r = 0.60), total cholesterol (r = 0.36), midarm circumference (r = 0.50), and triceps skinfold (r = 0.37). The sensitivity and specificity of the full MNA score (< 17) for hypoalbuminemia were 0.810 and 0.860, respectively. With a cutoff point lower than 18, sensitivity and specificity hypoalbuminemia were 0.857 and 0.815, respectively. Using a short-form MNA score 12 and higher as normal, its sensitivity and specificity for predicting undernutrition were 0.859 and 0.840, respectively.The full and short forms of the MNA were useful tools to identify elderly Japanese patients with malnutrition or risk of malnutrition. However, the full MNA cutoff point for malnutrition should be modulated for this population.