There is increasing interest in using combinations of two or more disease modifying anti-rheumatic drugs to treat rheumatoid arthritis. The use of such combinations is increasing in routine clinical practice. We have identified 18 well-conducted, randomised controlled trials of the use of combinations of disease modifying drugs, and a number of open studies that provide helpful supportive information. The 18 trials involved 2221 patients. Two trials reported strongly positive results, six reported moderately positive results and ten gave largely negative results. The combination of methotrexate, sulfasalazine and hydroxychloroquine appears to be effective with an acceptable level of adverse effects. There is also evidence that the combination of methotrexate and cyclosporin is advantageous. With both combinations, there appears to be further advantages from using corticosteroids in addition to the combination, although the evidence for this is incomplete. The use of other combinations is of less value, and in particular combinations involving parenteral gold, penicillamine and azathioprine are best avoided. Finally, there is growing evidence from randomised trials that the combination of anti-tumour necrosis factor (TNF) therapy with methotrexate is effective and well tolerated. We have identified four randomised controlled trials of the use of combinations of anti-TNF with methotrexate that all reported results favouring this combination. There is insufficient evidence to support the use of other combinations involving immunotherapies at the present time.