OBJECTIVES: There are 5 anti-TNF agents licensed for the treatment of RA in Ireland; adalimumab, certolizumab pegol, etanercept, golimumab and infliximab.  Reimbursement agencies have issued mixed approvals for their use through either a multiple assessment process or a single technology appraisal process.  Significant uncertainties have been identified through these assessments.  A multiple treatment comparison of these agents, focusing on the uncertainties previously highlighted, in the Irish health care setting is the focus of this study METHODS: The Birmingham Rheumatoid Arthritis Model 2009 was used to estimate the cost effectiveness of anti-TNF agents in patients with established RA who were non-responders to methotrexate.  The perspective taken is that of the Irish Healthcare Payer.  Evidence synthesis of HAQ data (via a multiplier), long and short term survival data was performed in WinBUGs and used to inform the effect parameter in the model.  Irish cost data was applied.  Utility mapping between HAQ, EQ-5D (revised scoring) and SF-6D was used to model utility gain.  Probabilistic analysis was performed. RESULTS: Incremental cost effectiveness ratios were calculated for each anti-TNF in comparison to methotrexate.  Infliximab is dominated in almost all scenarios, being more costly and less effective.  Etanercept and adalimumab are the most effective options and most costly.  Golimumab appears less effective and less costly.  Sensitivity analysis around utility measure and HAQ improvement assumptions indicate that these are key drivers to the estimates. CONCLUSIONS: This CUA focuses on novel methods for data inputs estimation.  Both Bayesian and frequentist methods are employed in order to validate the sub models used.  The outputs from the cost utility model show that etanercept and adalimumab appear most effective on the cost effectiveness plane.  However they are also more costly than both golimumab and certolizumab pegol.