Purpose: This study was done to examine relevance of the Patient-Generated Subjective Global Assessment (PG-SGA) to assess nutritional status of older hospitalized patients. Methods: This methodology study was performed between April 1 and August 30, 2010, in a tertiary hospital geriatric ward in Seoul. Data on demographics, disease-related factors, and nutritional factors were identified using a structured research form. The nutritional status was examined using the Geriatric Nutritional Risk Index (GNRI) score, body mass index, total lymphocyte counts, hemoglobin, albumin, and cholesterol at admission. Accuracy of the nutritional assessment by PG-SGA and GNRI, using a contingency table and ROC curve was compared with other nutritional parameters of undernourished and well-nourished patients as classified by the PG-SGA. Results: M±SD age of the patients was 72±7.6 years, and 43 (59.7%) were female. Compared with the GNRI, the PG-SGA showed 90% sensitivity, 26% specificity, and 61% positive and 67% negative predictive values, and area under ROC curve (AUC) was 0.64. None of the nutritional parameters differed between the undernourished and well-nourished patients (all p>.05). Conclusion: Use of the PG-SGA alone is not valid for nutritional assessment of acutely ill hospitalized older patients, and should only be used together with other objective nutritional parameters.