BACKGROUND Non-steroidal antiinflammatory drugs are extensively prescribed for the symptomatic treatment of osteoarthritis (OA). This study compared the efficacy and safety of nabumetone with aceclofenac in patients with active knee osteoarthritis. METHODS Phase IV, multicentre, prospective, open-label, parallel-group, randomized, three-month treatment, clinical trial, of nabumetone (1-2 g once a day) vs aceclofenac (100 mg two times each day). The main efficacy variable was intensity of pain, evaluated on a Visual Analogic Scale (VAS). In addition: Gonarthrosis Severity Index (GSI)-baseline score > 5 and < 17-, patient and physician Global Assessment of OA Activity, patient and physician Global Assessment of OA progression were assessed. Adverse events incidence and severity were assessed. Three study visits were made on a monthly basis. The main statistical analysis was by intention to treat (ITT). RESULTS In 12 centres 274 patients were recruited (137/group), aged 62.5 +/- 8.4 years, of whom 189 completed the trial (97 nabumetone and 92 aceclofenac). 90% were female, and 83% of the patients had both knees affected. In both groups the intensity of pain at month 3 improved significantly (p < 0.001) with respect to baseline: VAS of the nabumetone group: 6.1 +/- 1.9 to 4.3 +/- 2.8 cm; VAS of the aceclofenac group: 6.1 +/- 1.9 to 4.4 +/- 2.7 cm. There were no significant differences in any of the intergroup comparisons made. 15 nabumetone-treated and 23 aceclofenac-treated patients withdrew from the study due to moderate to severe adverse events. Gastrointestinal adverse events were the reason for withdrawal in 7 nabumetone and 14 aceclofenac-treated patients. CONCLUSIONS The efficacy and safety of nabumetone are similar to those of aceclofenac in the treatment of knee osteoarthritis.