We compared the benefits and risks of combinations of long acting antirheumatic drugs with those of the same drugs used singly in the treatment of rheumatoid arthritis. We searched the literature through MEDLINE (1966-89), Index Medicus (1956-65), Excerpta Medica (1982-89), Science Citation Index (1982-89), and bibliographic review of located articles. Of a total of 341 citations, we identified 7 prospective trials that specifically addressed the stated purpose. We independently assessed the quality of the selected trials, using published methodological criteria and summarized the effect of treatment on arthritis activity and the incidence of side effects. The trials we evaluated tested various drug combinations. Because of deficiencies in methods and reporting, only 3 trials had sufficient quality to yield strong or moderately strong evidence. None conclusively demonstrated benefit of a drug combination: 2 suggested such benefit, including 1 also suggesting increased toxicity; the 3rd suggested only increased toxicity. The other 4 trials yielded weak evidence to support both increased efficacy and toxicity. The advantages of any antirheumatic drug combination remain unproven. Because these advantages are likely to be modest, they can only be shown in rigorously designed trials enrolling large numbers of patients. Methods and reporting of antirheumatic drug trials should be standardized to allow combining of study results.