Aim: In a predefined analysis, data were pooled from eight blinded, randomized, controlled trials, and separately from three long-term, open-label trials to determine the rate of upper gastrointestinal ulcer complications with the cyclo-oxygenase-2 selective inhibitor, valdecoxib, vs. non-selective non-steroidal anti-inflammatory drugs. Methods: In randomized, controlled trials, 7434 osteoarthritis and rheumatoid arthritis patients received placebo (n = 973), valdecoxib 5-80 mg daily (n = 4362), or a non-selective non-steroidal anti-inflammatory drug (naproxen, ibuprofen or diclofenac; n = 2099) for 12-26 weeks. In long-term, open-label trials, 2871 patients received valdecoxib 10-80 mg daily for up to 1 year. All potential events were reviewed by a blinded, independent review committee based on a priori definitions of ulcer complications (perforations, obstructions, bleeds). Results: In randomized, controlled trials, 19 of 955 potential events were adjudicated to be ulcer complications. Valdecoxib was associated with a significantly lower ulcer complication rate than non-selective nonsteroidal anti-inflammatory drugs (0.68% vs. 1.96%, all patients; 0.29% vs. 2.08%, non-aspirin users; P < 0.05). In long-term, open-label trials, seven of 310 potential events were adjudicated to be ulcer complications; the annualized incidence for valdecoxib was 0.39% (seven of 1791 patient-years) for all patients and 0.2% (three of 1472 patient-years) for non-aspirin users. Conclusions: Valdecoxib, including above recommended doses, is associated with a significantly lower rate of upper gastrointestinal ulcer complications than therapeutic doses of non-selective non-steroidal anti-inflammatory drugs.