Abstract The performance of the WRF mesoscale model in the wind simulation and wind energy estimates was assessed and evaluated under different initial and boundary forcing conditions. Due to the continuous evolution and progress in the development of reanalyses datasets, this work aims to compare an older, yet widely used, reanalysis (the NCEP-R2) with three recently released reanalyses datasets that represent the new generation of this type of data (ERA-Interim, NASA-MERRA and NCEP-CFSR). Due to its intensive use in wind energy assessment studies, the NCEP-GFS and NCEP-FNL analysis were also used to drive WRF and its results compared to those of the simulations driven by reanalyses. Six different WRF simulations were conducted and their results compared to measured wind data collected at thirteen wind measuring stations located in Portugal in areas of high wind energy potential. Based on the analysis and results presented in this work, it can be concluded that the new generation reanalyses are able to provide a considerable improvement in wind simulation when compared to the older reanalyses. Among all the initial and boundary conditions datasets tested here, ERA-Interim reanalysis is the one that likely provides the most realistic initial and boundary data, providing the best estimates of the local wind regimes and potential wind energy production. The NCEP-GFS and NCEP-FNL analyses seem to be the best alternatives to ERA-Interim, showing better results than all the other reanalyses datasets here tested, and can therefore be considered as valid alternatives to ERA-Interim, in particular for cases where reliable forcing data is needed for real-time applications due to its fast availability.