Modern treatment of RA includes the use of biologics. Their cost is high and comparison between different treatment strategies is needed. Direct medical costs of RA in France were evaluated based on expert opinion. Then, simulation-decision analytical models were developed to assess four biologic treatment sequences over 2 years in patients failing to respond to at least one anti-TNF agent. Effectiveness was expressed in theoretical expected number of days (TEND) in remission or low disease activity [low disease activity score (LDAS)] based on DAS-28 scores. Direct medical costs of RA in France (excluding the cost of biologics) were estimated at euro 905 (s.d. 263) for 6 months and euro 696 (s.d. 240) for each subsequent 6 months (P < 0.001) for patients achieving LDAS and euro 1215 for 6 months (s.d. 405) for patients not achieving LDAS. Based on LDAS criteria, using abatacept after an inadequate response to the first anti-TNF agent (etanercept) appeared significantly (P < 0.01) more efficacious over a 2-year period (102 TEND) compared with using rituximab at a 6-month re-treatment interval (82 TEND). Mean cost-effectiveness ratios showed significantly lower costs (P < 0.01) per TEND with abatacept as second biologic agent (euro 278) compared with rituximab (euro 303). After an inadequate response to two anti-TNF agents, using abatacept also appeared significantly more efficacious than an anti-TNF agent (P < 0.01). All comparisons were confirmed when using remission criteria instead of LDAS. Advanced simulation models based on clinical evidence and medical practice appear to be a promising approach for comparing cost-effectiveness of biologic strategies in RA.