Resolution plays a major role in the accuracy of wave simulations, and its definition is an important step of wave modelling. It should be chosen taking into consideration the specific goals of each single study, balancing between the need of local space/time skill, the known limits of the employed wind forcing and of the implemented wave physics, and the constraints represented by the available computation resources. In this study we analyzed the effect of wind and wave model resolution on global-scale wave simulations, employing WAVEWATCHIII and wind forcing from CFSR reanalysis and a downscaled CMIP5 model. The resolution of both the model and wind forcing were varied between 0.4 and 1.5°. Decreasing wind resolution alone, without recalibrating the model, leads to negative bias in the simulated significant wave height due to smoother forcing wind peaks. This effect is more than compensated if the wave resolution is reduced, due to the numerical tendency to overestimate wave growth. Improving subscale modelling, which plays a major role in the skill of low-resolution models, leads to significantly better performances in many areas. Although the skill at a resolution of 0.4° is slightly better, the performance at 0.75° and 1.5° is comparable, both in terms of mean and extreme significant wave height. Finally, the impact of resolution (0.75° and 1.5°) on projected waves under a high-warming climate change scenario is not significant. Our results suggest that, unless substantial improvement of atmosphere-ocean dynamics is introduced in the model, lower resolution simulations can be used without major loss of information in studies that focus on global patterns and tendencies, or on overall conditions in large areas.