OBJECTIVES: To build a Markov Model that is able to assess the cost-effectiveness of adding NAT to the HIV screening strategy at the Brazilian public health system. METHODS: A mathematical model was made of the transfusion chain from donors to recipients of blood in Brazil. The annual number of avoided HIV transmissions was estimated with the window-period incidence model. The natural history of the whole blood receptors is described by a Markov model. RESULTS: The incremental cost-effectiveness ratio - ICER of using NAT instead of ELISA concurrently to another ELISA in Brazilian public health system is R$666.493,56 per QALY. The Brazilian gross domestic product per capita is R$18.315,50. CONCLUSIONS: The Markov Model built is consistent and shows that, despite the narrower window period, at current costs, NAT is not cost-effective for HIV screening of donated blood in Brazil. Given that the NAT kit is already being produced in-house, the price per kit can be adjusted to achieve a better ICER.