The risk of significant injury to the gastrointestinal (GI) tract from nonsteroidal anti-inflammatory drugs (NSAIDs) has been well established. The Veterans' Affairs (VA) implemented treatment criteria for the use of NSAIDs including the new class of drugs, cyclooxygenase -2 (COX-2) inhibitors. These criteria utilize a self-administered Gastrointestinal Risk Assessment Tool (GI Score), developed from the Arthritis, Rheumatism, and Aging Medical Information System (ARAMIS) database, to assess risk. This tool generates a composite score used to predict the 1-year risk level, level 1 (no risk) to level 4 (substantial risk), for the potential of an NSAID-associated GI event. OBJECTIVES: The purpose of this study was to assess the risk level and the level of implementation of the VA criteria. METHODS: The GI score was used to assess the patient's risk level calculated on the basis of data from VA demographic, prescription, hospitalization, clinic visits, and active problem lists databases. Current therapy was compared to criteria-based therapy to assess level of implementation. RESULTS: There were 7,625 NSAID users in the New Mexico VA Healthcare System: 86 previous hospitalized GI event patients, 100 concurrent warfarin therapy patients, 223 corticosteroid therapy patients, and 205 rheumatoid arthritis patients. Thirty-six percent of the VA patients were over the age of sixty-five. The most commonly prescribed NSAIDs were ibuprofen (47%), naproxen (22%) and etodolac (14%). The calculated risk level indicated that 29% of patients had substantial (4%) or significant (25%) risk for a potential GI event, of which 34% were prescribed criteria-based therapy (salsalate, non-selective NSAID plus proton pump inhibitor (PPI), high-dose famotidine or misoprostol, or COX-2 inhibitor). CONCLUSIONS: In this study, salsalate, non-selective NSAIDs with PPI, high-dose famotidine or misoprostol, and COX-2 inhibitor are under utilized in patients at high risk for a GI event.