Significant opportunities for patient safety are potentially achievable with the introduction of the cycloxygenase-2 (COX-2) inhibitors. Unfortunately, constrained health-care budgets may limit their use. OBJECTIVE: To describe a NSAID therapeutic portfolio that maximizes medical benefits, but does not increase existing health-care costs. METHODS: A meta-heuristic assisted optimization of a typical HMO’s NSAID portfolio after the introduction of celecoxib 200 mg QD was conducted using linear programming techniques. Current NSAID?antiulcerant usage and the cost of managing NSAID related gastropathy was derived from a large HMO. Background antiulcerant usage was established from an equivalent nonNSAID exposed arthritis population. The probabilities of NSAID related gastrointestinal events were derived from clinical trials and the literature. Decreased hospitalized gastrointestinal events without increasing the overall health budget were defined as the optimization criteria. Antiulcerant usage was not allowed to decrease below the background rate RESULTS: Assuming an average risk of NSAID-induced GI toxicity and 150 days/year of therapy, the average cost/patient/year for NSAID therapy and related adverse events would be $366. If celecoxib is introduced into the health-care system, 74% of patients could be switched to the agent without increasing the overall budget, provided that the remaining patients were distributed as follows: 13% generic ibuprofen; 9% generic NSAID+H2 blockers or prostaglandins; and 4% generic NSAID+PPI. During a one year period, this distribution of agents would result in 29 fewer physician visits and two less hospitalization per 1000 patients. CONCLUSIONS: These results indicate that by optimizing the distribution of NSAIDs utilized within a HMO, a significant number of osteoarthritic patients can receive safer COX-2 specific compounds such as celecoxib without increasing the total budget.