Background: Inflammatory mechanisms have been implicated in the pathogenesis of Alzheimer's disease (AD) and may be mediated via the cyclo-oxygenase-2 enzyme. This study sought to evaluate the effect of rofecoxib, a nonsteroidal anti-inflammatory drug that selectively inhibits cyclo-oxygenase-2, in slowing the progression of dementia in patients with established AD. Methods: A double-blinded, multicenter trial was conducted in which 692 patients with mild or moderate AD aged 50 years or older were randomly assigned to receive 25 mg rofecoxib or placebo daily for 12 months. The key efficacy measures were mean change from baseline at month 12 on the cognitive subscale of the AD Assessment Scale (ADAS-cog) and score on the Clinician's Interview Based Impression of Change with caregiver input (CIBIC+). Results: Four hundred eighty-one patients (70%) completed assessments and remained on treatment at 12 months. No significant differences between treatments were found on the mean change from baseline error score for the ADAS-cog (rofecoxib = 4.84; placebo = 5.44; difference = -0.60) or mean score on the CIBIC+ (rofecoxib = 4.90; placebo = 4.87; difference = 0.03) over 12 months. This result persisted after adjusting for severity of dementia at baseline, presence of the APOE-ε4 allele, and donepezil use. Secondary analyses did not reveal any significant differences on any other measures. Conclusion: The failure of selective cyclo-oxygenase-2 inhibition to slow the progression of AD may indicate either that the disease process is too advanced to modify in patients with established dementia or that cyclo-oxygenase-2 does not play a significant role in the pathogenesis of the disorder.