OBJECTIVES: On September 30, 2004, rofecoxib—a drug used to treat inflammation—was voluntarily withdrawn from the US market due to mounting evidence of increased cardiovascular (CV) risk. This study sought to assess movement in the number of clinically appropriate (CAP) cyclo-oxygenase (COX)-2 prescriptions dispensed after the rofecoxib withdrawal; and whether patients with CAP profiles demonstrated different post-withdrawal switching behavior or incurred different levels of healthcare costs. METHODS: Administrative claims data from four regional health plans were used. To examine differences in clinically appropriate prescribing pre- and post-rofecoxib withdrawal, a generalized estimating equations approach was used. RESULTS: Based on a total of 317,762 prescriptions, this study found that CAP prescribing rates were lower after news of the increased drug risk, as were the total number of COX-2 prescriptions that were clinically inappropriate. Patients with CAP risk were more likely to exhibit switching behavior by filling prescriptions for nonselective (ns)NSAIDs or discontinuing use of all NSAIDs. Healthcare costs incurred in the year following the rofecoxib withdrawal were lower for patients who were prescribed COX-2s appropriately. CONCLUSION: These findings suggest that the widespread coverage of the rofecoxib withdrawal, which led to heightened awareness of the drug's risk affected consumer demand for COX-2s and may have had unintended consequences in reducing the number of patients who “should have been” receiving these medications according to clinical risk. Because patients and physicians were forced to make a medical decision under risk, increased information of the benefit-risk tradeoff may have been helpful and may have contributed to greater consumer welfare.