OBJECTIVES: Although second generation Tyrosine kinase inhibitor (TKIs) nilotinib and dasatinib are associated with higher rates of sustained deep molecular response (SDMR) compared to imatinib, the later remains as frontline therapy for chronic myeloid leukaemia (CML) in Brazil due to lower prices of generic formulations. However, as treatment free remission (TFR) became feasible for CML patients, second generation TKIs might be a valuable option for frontline therapy since SDMR is the main eligibility criterion for treatment suspension. This study compared the cost-effectiveness of first and second generation TKIs as frontline CML treatment considering TRF as a possibility. METHODS: We constructed a Markov model to compare the ten year cost-effectiveness of imatinib frontline therapy versus nilotinib or dasatinib from a Brazilian third payer perspective, assuming 5% annual discounting. The model included six health states: first line treatment, second line treatment, TFR, treatment resume, disease progression and death. Transition probabilities were obtained from eligible studies. Only direct medical costs were considered including those related to drugs acquisition, medical visits and laboratory tests (exchange rate 1 BRL = 4 USD). The main outcome was progression-free life-years (PF-LY). We assessed different price scenarios for second generation TKIs. RESULTS: Nilotinib-first strategy offered 0.85 additional PF-LY with an ICER of US$ 29,541.98/PF-LY compared to imatinib. Dasatinib frontline was the most costly strategy with an ICER of US$ 81,355.07/PF-LY. These results were robust to probabilistic sensitivity analysis. The scenarios analysis showed that a 25% reduction in nilotinib price would make it the dominant strategy. CONCLUSIONS: Although TFR was included as a health state in the model, imatinib remained as the most economically advantageous frontline CML treatment, mainly because of the low price of its generic formulations. However, a decrease in the price of nilotinib could significantly change this scenario.