COX-2 specific inhibitors have demonstrated significant safety advantages and comparable efficacy in osteoarthritis (OA) compared with conventional nonsteroidal anti-inflammatory drugs (NSAIDs), but no direct comparative trials between COX-2 specific inhibitors have been published. In this double-blind, placebo-controlled, parallel group, multicenter study, 182 patients (≥40 years old) with OA of the knee were randomly assigned to treatment with celecoxib 200 mg q.d. (n = 63), rofecoxib 25 mg q.d. (n = 59), or placebo (n = 60) for 6 weeks. Arthritis assessments were performed at baseline and Weeks 3 and 6, or at early termination. At Week 6, celecoxib and rofecoxib treatment resulted in similar mean changes from baseline (p> 0.55) in arthritis pain visual analogue scale, patient's global assessment, and total score for WOMAC; all changes were superior to placebo (p< 0.05). In the patient's global assessment of arthritis pain at Week 6, 79% of celecoxib-treated and 78% of rofecoxib-treated patients improved by ≥1 grade, compared with 50% of placebo patients (celecoxib,p= 0.025; rofecoxib,p= 0.020). Adverse event incidences were similar among the active comparators; however, celecoxib-treated patients had significantly fewer adverse gastrointestinal symptoms compared with rofecoxib-treated patients, which suggests that celecoxib may have a better gastrointestinal tolerability profile than rofecoxib at these doses. Adverse events that prompted withdrawal occurred in fewer than 7% of patients, and the overall incidences were similar between the active agents. Once-daily doses of celecoxib 200 mg and rofecoxib 25 mg offer comparable efficacy and are an effective alternative to conventional NSAIDs in the management of OA.