This multicenter, 6-week, double-blind, placebo-controlled, parallel-group study compared the efficacy and safety of oxaprozin 1200 mg once daily with that of nabumetone 1000 mg once daily in patients with moderate-to-severe osteoarthritis (OA) of the knee. To be eligible, patients had to experience a flare of OA within 2 weeks of discontinuing their usual OA medication (nonsteroidal anti-inflammatory drug or analgesic). Eligible patients were assessed at baseline and then randomized to receive oxaprozin (n = 109), nabumetone (n = 110), or placebo (n = 109). Efficacy assessments were performed at weeks 1, 2, 4, and 6. Primary efficacy variables included knee pain on weight bearing, knee pain on motion, and patient's and physician's global assessments of OA. Secondary efficacy variables included pain intensity, time to walk 50 feet, and duration of morning stiffness. Safety was evaluated by use of routine laboratory analyses; physical examination at screening, baseline, and week 6 (or study termination); assessment of symptoms at baseline and at each visit; and testing stools for occult blood at screening and between week 4 and the final visit. Adverse events were monitored throughout the study. Between-group differences in efficacy variables were evident by week 1. The mean change in improvement from baseline with oxaprozin compared with placebo was statistically significant in favor of oxaprozin at weeks 1, 2, 4, and 6 for all primary efficacy variables. The mean change in improvement from baseline with nabumetone compared with placebo, however, was statistically significant only at week 1 for knee pain on motion, patient's global assessment, and physician's global assessment. The mean change in improvement from baseline was statistically significant (P < or = 0.035) in favor of oxaprozin versus nabumetone at weeks 2 and 6 for all four primary efficacy variables and also at week 4 for knee pain on motion. The incidence of adverse clinical events between treatment groups was not statistically significant. However, nine oxaprozin-treated patients had asymptomatic liver enzyme elevations reported as adverse events. Four of these patients had reversible elevations of aspartate aminotransferase and alanine aminotransferase greater than three times the upper limit of normal range (P < 0.05); two of these patients were taking other medications known to induce liver enzyme abnormalities. The study showed that oxaprozin 1200 mg once daily was statistically significantly more efficacious than nabumetone 1000 mg once daily for the treatment of patients with moderate-to-severe OA of the knee. Both drugs were clinically well tolerated.