Simulated wind fields from regional climate models (RCMs) are increasingly used as a surrogate for observations which are costly and prone to homogeneity deficiencies. Compounding the problem, a lack of reliable observations makes the validation of the simulated wind fields a non trivial exercise. Whilst the literature shows that RCMs tend to underestimate strong winds over land these investigations mainly relied on comparisons with near surface measurements and extrapolated model wind fields. In this study a new approach is proposed using measurements from high towers and a robust validation process. Tower height wind data are smoother and thus more representative of regional winds. As benefit this approach circumvents the need to extrapolate simulated wind fields. The performance of two models using different downscaling techniques is evaluated. the influence of the boundary conditions on the simulation of wind statistics is investigated. Both models demonstrate a reasonable performance over flat homogeneous terrain and deficiencies over complex terrain, such as the Upper Rhine Valley, due to a too coarse spatial resolution (~50 km). When the spatial resolution is increased to 10 and 20 km respectively a benefit is found for the simulation of the wind direction only. A sensitivity analysis shows major deviations of international land cover data. A time series analysis of dynamically downscaled simulations is conducted. While the annual cycle and the interannual variability are well simulated, the models are less effective at simulating small scale fluctuations and the diurnal cycle. The hypothesis that strong winds are underestimated by RCMs is supported by means of a storm analysis. Only two-thirds of the observed storms are simulated by the model using a spectral nudging approach. In addition “False Alarms” are simulated, which are not detected in the observations. A trend analysis over the period 1961 - 2000 is conducted for two RCM simulations and their driving reanalysis. the RCMs generally reproduce the trend pattern of the driving fields. On regional scales, deviations occur due to their higher resolution and the expected added value for complex terrain. A piecewise trend analysis reveals two dominant trend patterns. These can be linked to a positive NAo index and a northward shift of the North Atlantic storm track until 1990 and a southward shift afterwards.