OBJECTIVES: To provide a review of the decision analytic models used in the economic evaluations of Tumor Necrosis Factor (TNF) inhibitors (adalimumab, etanercept, and infliximab) in Rheumatoid Arthritis (RA) and to address some important issues surrounding the choice of such analytic modeling techniques in these economic analyses. METHODS: A systematic literature search was conducted by one researcher among publications in peer-reviewed journals from January, 1996 to October, 2004 through MEDLINE and EMBASE databases to identify studies that used decision analysis models to evaluate cost-effectiveness of TNF inhibitors in patients with RA. All the studies of TNF inhibitors in patients other than RA, and those conducted in children were excluded from the review. RESULTS: The systematic literature search identified 28 articles. Of these only ten studies fulfilled the inclusion criteria and were included in the review process. These ten studies used different decision analysis models, which are listed as follows: decision trees (two studies), Markov model (four studies), Monte Carlo Simulation (two studies), and Discrete Event Simulation (two studies). Since the models vary in complexity, the choice of these modeling techniques depends on the course of the disease, impact of the drugs, and the availability of data. The results of most of the studies indicate that all three TNF inhibitors are cost-effective compared with traditional agents and have cost-effectiveness ratios of less than $50,000/QALY gained. However, one study reports the cost-effectiveness ratio of more than $100,000/QALY for etanercept and infliximab. CONCLUSION: Based on the results derived from different modeling techniques, it would seem that all methods provide useful techniques for economic evaluations of TNF inhibitors. However, to increase the confidence of the physicians and payers, key issues such as validity of the models, transparency during construction of the models, quality of data sources, and handling of uncertainty need to be resolved.