Ocean surface wind data derived from several QuikSCAT products and the Cross-Calibrated Multi-Platform (CCMP) project were compared to wind speed and direction measurements, in order to assess which one of these databases has higher accuracy and ability to describe the local wind regime characteristics. For this, data from QuikSCAT (swath data from L2B 25 km and L2B 12.5 km slice composites, together with gridded L3) and CCMP were compared with measurements taken from five buoys located along the Iberian Peninsula coast. The results presented in this work show that QuikSCAT products have their strength in representing the temporal variability of the wind speed (higher correlation coefficients and lower RMSE and STDE) and the mean state of the wind direction (lower biases). Although no major differences were detected among QuikSCAT products, the high-resolution database (12.5 km) was the one with the best overall scores. However, CCMP is able to bring significant improvements in terms of wind direction temporal variability and wind speed mean state. CCMP also showed its capability to partially mitigate some of QuikSCAT's known problems, mainly those related to QuikSCAT systematic tendency to overestimate the wind speed and land masking effects. In addition, CCMP consists in a gridded dataset with a higher temporal sampling and complete data availability when compared to QuikSCAT, allowing this database to be clearly the one with a better ability to characterize the wind regimes measured by the buoys in terms of wind speed frequency distributions. These features can render CCMP an interesting ocean wind database for offshore wind energy assessment studies, where wind speed mean state accuracy plays the key role, and also for meteorological, oceanic and climate modelling applications where gridded wind data with good temporal sampling and data availability is vital to force numerical simulation models. (c) 2013 Elsevier Inc. All rights reserved.