Tumour necrosis factor alpha (TNFalpha) inhibitors are frequently used to treat rheumatoid arthritis, but whether use of a different TNFalpha inhibitor can improve patient response is unknown. We assess the efficacy and safety of the TNFalpha inhibitor golimumab in patients with active rheumatoid arthritis who had previously received one or more TNFalpha inhibitors. 461 patients with active rheumatoid arthritis from 82 sites in 10 countries were randomly allocated by interactive voice response system, stratified by study site and methotrexate use, to receive subcutaneous injections of placebo (n=155), 50 mg golimumab (n=153), or 100 mg golimumab (n=153) every 4 weeks between Feb 21, 2006, and Sept 26, 2007. Allocation was double-blind. Eligible patients had been treated with at least one dose of a TNFalpha inhibitor previously. Patients continued stable doses of methotrexate, sulfasalazine, hydroxychloroquine, oral corticosteroids, and non-steroidal anti-inflammatory drugs. The primary endpoint was achievement at week 14 of 20% or higher improvement in American College of Rheumatology criteria for assessment of rheumatoid arthritis (ACR20). At week 16, patients who had less than 20% improvement in tender and swollen joint counts were given rescue therapy and changed treatment from placebo to 50 mg golimumab, or from 50 mg to 100 mg golimumab. Drug efficacy was assessed by intention to treat and safety was assessed according to the study drug given. This study is registered with ClinicalTrials.gov, number NCT00299546. Patients had discontinued previous TNFalpha inhibitors because of lack of effectiveness (269 [58%] patients) or reasons unrelated to effectiveness (246 [53%] patients), such as intolerance and accessibility issues. Patients had active disease, which was indicated by a median of 14.0 (IQR 9.0-22.0) swollen and 26.0 (16.0-41.0) tender joints for the whole group. 28 (18%) patients on placebo, 54 (35%) patients on 50 mg golimumab (odds ratio 2.5 [95% CI 1.5-4.2], p=0.0006), and 58 (38%) patients on 100 mg golimumab (2.8 [1.6-4.7], p=0.0001) achieved ACR20 at week 14. Two patients were never treated, and 57 patients did not complete the study because of adverse events, unsatisfactory treatment effect, loss to follow-up, death, or other reasons. 155 patients on placebo, 153 on 50 mg golimumab, and 153 on 100 mg golimumab were assessed for drug efficacy. For weeks 1-16, serious adverse events were recorded in 11 (7%) patients on placebo, 8 (5%) on 50 mg golimumab, and 4 (3%) on 100 mg golimumab. For weeks 1-24, after some patients were given rescue therapy, serious adverse events were recorded in 15 (10%) patients on placebo, 14 (5%) on 50 mg golimumab, and 8 (4%) on 100 mg golimumab. Golimumab reduced the signs and symptoms of rheumatoid arthritis in patients with active disease who had previously received one or more TNFalpha inhibitors. Centocor Research and Development and Schering-Plough Research Institute.