BACKGROUND:New treatment strategies for early rheumatoid arthritis are evolving rapidly. We aimed to compare addition of conventional disease-modifying antirheumatic drugs (sulfasalazine and hydroxychloroquine) with addition of a tumour necrosis factor antagonist (infliximab) to methotrexate in patients with early rheumatoid arthritis.METHODS:We undertook a randomised trial in 15 rheumatology units in Sweden. We enrolled patients with early rheumatoid arthritis (symptom duration <1 year) and administered methotrexate (up to 20 mg per week). After 3-4 months, those who had not achieved low disease activity but who could tolerate methotrexate were randomly allocated by computer addition of either sulfasalazine and hydroxychloroquine or infliximab. Primary outcome was achievement of a good response according to European League Against Rheumatism (EULAR) criteria at 12 months. Patients were followed up to 24 months; here, we present findings at 12 months. Analysis was by intention to treat and we used non-responder imputation. The Swefot (Swedish Pharmacotherapy) study is registered in the WHO database at the Karolinska University Hospital, number CT20080004.FINDINGS:487 patients were initially enrolled. Of 258 who had not achieved low disease activity with methotrexate, 130 were allocated sulfasalazine and hydroxychloroquine and 128 were assigned infliximab. 32 of 130 (25%) patients allocated sulfasalazine and hydroxychloroquine achieved the primary outcome compared with 50 of 128 (39%) assigned infliximab (risk ratio 1.59 [95% CI 1.10-2.30], p=0.0160). Adverse events were balanced fairly well between the two groups and accorded with known adverse events of the drugs used. No deaths occurred in either group.INTERPRETATION:In patients with early rheumatoid arthritis in whom methotrexate treatment failed, addition of a tumour necrosis factor antagonist to methotrexate monotherapy is clinically superior to addition of conventional disease-modifying antirheumatic drugs.FUNDING:Swedish Rheumatism Association, Schering-Plough.