Steady-state inhibitory activity of rofecoxib (Vioxx) on COX-2 versus COX-1 was compared with that of commonly used nonsteroidal anti-inflammatory drugs (NSAIDs) in 76 healthy volunteers randomized to placebo, rofecoxib 12.5 mg qd, rofecoxib 25 mg qd, diclofenac 50 mg tid, ibuprofen 800 mg tid, sodium naproxen 550 mg bid, or meloxicam 15 mg qd. All of these doses include the high end of the approved clinical dose range. Ex vivo whole-blood assays were used to determine the effect on COX-2 and COX-1 activity, respectively. Urinary prostanoids were also measured. Mean inhibition of COX-2 (measured as the weighted average inhibition [WAI] of lipopolysaccharide [LPS]-induced PGE2 generation over 8 hours on day 6 vs. baseline) was -2.4%, 66.7%, 69.2%, 77.5%, 93.9%, 71.4%, and 71.5% for placebo, rofecoxib 12.5 mg, rofecoxib 25 mg, meloxicam, diclofenac, ibuprofen, and naproxen, respectively. Corresponding values for mean inhibition of COX-1 (measured as TXB2 generation in clotting whole blood) were -5.15%, 7.98%, 6.65%, 53.3%, 49.5%, 88.7%, and 94.9%. Rofecoxib had no significant effect on urinary excretion of 11-dehydro TXB2, a COX-1-derived product. These data support the contention that rofecoxib is the only drug of the regimens tested that uniquely inhibits COX-2 without affecting COX-1.