To determine if any nonsteroidal antiinflammatory drug (NSAID) was superior in the treatment of rheumatoid arthritis, duration of use of each drug was employed as a measure of combined efficacy and tolerability. Duration was treated as survival data and a proportional hazards model utilized to adjust for differences in disease severity and concomitant antirheumatic therapy. One hundred and sixteen patients took 188 courses of nonsalicylate NSAID during the 3-year study period. The NSAID prescribed included naproxen, ibuprofen, sulindac, indomethacin, piroxicam, and tolmetin. Naproxen was used significantly longer than any other NSAID (p less than 0.001).