Previous studies of combinations of nonsteroidal drugs used in the treatment of rheumatoid arthritis (RA) have yielded conflicting results. We used standard methods to measure disease activity and high pressure liquid chromatography to measure plasma drug concentrations. We used doses of choline magnesium trisalicylate, adjusted to achieve therapeutic serum salicylate concentrations, and naproxen in a randomized, double‐blind, placebo‐controlled cross‐over study of full dose trisalicylate (CMT), full dose naproxen (N), full dose of both (CMT‐N), and half dose of both (cmt‐n) to examine their relative efficacy and toxicity in treating RA. CMT‐N was statistically superior to all other treatments in only 1 of 12 efficacy variables, but was equal to N and better than CMT or cmt‐n for 7 variables. There were minimal differences among treatments for the other 4 efficacy variables. The mean percentage difference for the efficacy variables between CMT‐N and N was 3%, between CMT‐N and CMT was 10.6%, and between CMT‐N and cmt‐n was 10.5%. Thirteen percent of patients manifested toxic reactions during the initial open dose‐adjustment salicylate run‐in phase. During the double‐blind phases of the study, CMT‐N was more toxic than N, CMT, or cmt‐n (7.5% versus 3.4%, 1.8%, and 3.7%, respectively). Tinnitus was more common when full‐dose CMT was used; N (N or CMT‐N) was associated with increased skin toxicity. Gastrointestinal complaints were equally common with all regimens. CMT‐N, although sometimes statistically superior to CMT, N, or cmt‐n, showed no clinically important additive or synergistic effect versus N or CMT alone.