Background and Objective: Nutritional assessment in the elderly is difficult and different from that performed in younger people. There are specific tools for that purpose, such as the Geriatric Nutritional Risk Index (GNRI). The study objective was to compare this index to the Nutritional Risk Index (NRI)., Materials and Methods: A retrospective, observational analytical study including 113 hospitalized patients over 75 years of age receiving nutritional support. Weight, height, age, length of stay, Mini Nutritional Assessment (MNA), nutrition type and duration, and occurrence of complications were collected. GNRI and NRI were calculated. Both indexes were compared to each other and with parameters measured., Results: Mean GNRI was 88.79 (SD: 13.1), mean NRI 79.96 (SD: 10.8), and mean MNA 17.49 (SD: 4.9). Complications occurred in 50.4% of patients, and 14% died. NRI and GNRI did not correlate with length of stay (R=0.136) or with length of nutrition (R=0.041). No significant correlation was seen between GNRI and complications, but a significant relationship was found with NRI. After stratification into surgical and medical patients, NRI was seen to be significantly related to complications in surgical patients only (p<0.05). GNRI was not related to complications in either surgical or medical patients., Conclusions: In hospitalized elderly patients, NRI is a better predictor of complications and may be more appropriate for assessing the risk of death than GNRI. GNRI underestimated nutritional risk as compared to NRI., (Copyright © 2010 SEEN. Published by Elsevier Espana. All rights reserved.)