OBJECTIVES: Conventional cost-effectiveness analysis are usually supported by clinical data collected in phase III RCTs (Drummond, 2005) and it is typically based on time-to-event data. However, ignoring mean quality-adjusted survival for multiple transitions may introduce some estimates bias. Multi-state models appeared to eliminate some limitations of Q-TWiST models such as the necessity for progressive states (Billingham, 1999), but the applicability of these models in heath economics is still unclear. The aim of this analysis is to compare the results of two modeling alternatives: conventional cost-effectiveness analysis and multi-state cost-effectiveness analysis taking capecitabine versus docetaxel in mBC in Brazil as a study case. METHODS: Conventional and multi-state survival analysis techniques were applied on O'Shaugnessy (2001) data to perform both cost-effectiveness analysis. For the conventional survival analysis, a Cox and parametric approach were used, and the best fit was chosen analyzing Cox-Snell and deviance residuals, as well as AIC and BIC criteria. For multi-state survival analysis, the model estimation was based on Gardiner (2006), and assumed as a three-stage non-homogeneous Markov process. For costs, the resource use was obtained from O'Shaugnessy trial and unit costs were derived from published lists such as CBHPM 2004, Simpro magazine and PROAHSA report. RESULTS: In conventional analysis the incremental cost-effectiveness ratio (ICER) was R$ 26.208 per QALY assuming log-logistic distributions which was the best fitted one, meanwhile in multi-state analysis, the incremental cost-effectiveness ratio was R$ 19.320 per QALY using weilbull distributions (Klein, 1994). PSA was performed using Monte-Carlo simulations that supported the robustness of the findings. CONCLUSION: Despite multi-state analysis demands more sophisticated statistical background, there is an evident superiority in multi-state estimates precision and accuracy when performing cost-effectiveness analysis. Regarding to capecitabine, it was observed that in both models its addition in this setting could be considered as a cost-effective alternative under the private payer perspective in Brazil.