Pressure coefficients (Cp) are a key parameter in the study of wind-driven natural ventilation in buildings. This paper reports the results of a research study aimed at evaluating the feasility of using CFD (Computational Fluid Dynamics) to obtain Cp data and also evaluating the importance of some of the aspects of the simulation setup in the results obtained. The following aspects are addressed: mesh independence, sensitivity to the wind profile and to the roughness of the domain floor. Values of y+ are discussed, and surface-averaged Cp values from the literature are compared to local Cp calculated by CFD. The building used in the simulations is a 5 storey, isolated building, typically adopted in social housing projects in Brazil. This paper concludes that CFD simulations are feasible and constitute an important alternative to obtain Cp values when compared to wind tunnel experiments. However, differences in the simulation setup may lead to significant deviations in the results, both in absolute values (up to ± 0.5) and in relative values (up to 50%). The mesh independence test has proven to be fundamental for a reliable simulation. The same applies for the inlet wind profile definition. Future work should focus on the validation of simulations.