This 6-week, multicenter, double-masked, placebo-controlled study compared the efficacy, tolerability, and safety of the recommended starting dose of oxaprozin (1200 mg/d) and a 1500-mg/d dose of nabumetone in the treatment of patients with moderate-to-severe osteoarthritis (OA) of the knee. A total of 347 patients with a mean age of 61.1 years were randomized to receive oxaprozin (116 patients), nobumetone (115 patients), or placebo (116 patients). Adults of either sex who were older than 18 years of age were eligible for entry into the study, if they had had OA of the knee for at least 6 months. Efficacy variables included knee pain on weight bearing, knee pain on motion, patients' and physicians' global assessments of OA, pain intensity as measured on a visual analog scale, and time to walk 50 feet as quickly as possible. Efficacy variables were assessed at baseline and at weeks 1, 2, 4, and 6. Between-group differences in efficacy variables were evident by week 1. Mean improvements were significantly greter with oxaprozin than with placebo for all efficacy variables at all time periods, except knee pain on motion at weeks 2 and 4 and time to walk 50 feet at weeks 1, 2, and 4. Mean improvements were significantly greater with nabumetone than with placebo for all efficacy variables at all time periods, except the following: knee pain on weight bearing at weeks 2, 4, and 6; knee pain on motion at weeks 2 and 4; patients' global assessment at week 4; and pain intensity as measured on a visual analog scale at weeks 2 and 4. There were, however, no significant differences between oxaprozin and nabumetone in any of these efficacy variables. Adverse events were reported by 83 (71.6%) patients who took exaprozin, by 80 (69.6%) patients who took nabumetone, and by 57 (49.1%) patients who took placebo. Adverse events were reported for significantly more patients taking oxaprozin or nabumetone than placebo. However, adverse events tended to be mild or moderate and rarely resulted in patients withdrawing from the study. Combined with the results of an earlier study, the results of this showed that a 1500-mg/d dose of nabumetone, which is higher than the recommended starting dose of 1000 mg/d, is required for efficacy equivalent to that of the recommended starting dose of oxaprozin, 1200 mg/d, in relieving the symptoms of OA. Thus nabumetone may require dosage titration from the recommended starting dose. Oxaprozin and nabumetone were found to have similar tolerability profiles, as shown by adverse-event monitoring and withdrawal rates, as well as clinically similar safety profiles, as demonstrated by physical examinations, hematologic and biochemical laboratory testing, hemoccult testing, and adverse-event monitoring and symptom assessment.