Many new nonsteroidal antiinflammatory agents are available for the treatment of rheumatoid arthritis. Although they all seem effective and reasonably safe, it is difficult to determine whether some are superior over others under real life conditions of use. Currently available study designs for answering this question were evaluated. Results demonstrated the value of treatment duration as a measure of the efficacy and safety of nonsteroidal antiinflammatory drugs. An earlier retrospective study using treatment duration is described and its results compared with those reported by other investigators. Treatment duration was subjected to survival analysis, and the Cox proportional hazards model was used to adjust for disease severity and concomitant treatments. One hundred sixteen patients with rheumatoid arthritis meeting American College of Rheumatology criteria received 188 courses of nonsteroidal antiinflammatory drugs over the four-year study period. Regardless of the nonsteroidal antiinflammatory agent used, treatment duration was longer in patients receiving steroid therapy, a second-line drug, or a narcotic analgesic. Treatment duration was not influenced by disease severity, the number of nonsteroidal antiinflammatory agents used, the number of previously used second-line drugs, the dosage, or the prescribing physician. Treatment duration was significantly (p < 0.001) longer for naproxen than for the other antiinflammatory agents (sulindac, piroxicam, indomethacin, tolmetin, and ibuprofen). This difference persisted after adjustment for concomitant use of prednisone, second-line drugs, or analgesics. Other studies support our findings, although they found no statistically significant differences, for a number of reasons. In our study, naproxen was used longer than any other nonsteroidal antiinflammatory drug. Continued work is needed to confirm our findings.