A two-layer boundary layer model is presented for regional assessment of seasonal-average surface wind climate from station data. In the surface layer the wind measurements are objectively corrected for imperfect station exposure, using gustiness-derived roughness values for each azimuth sector. This produces at 60 m a mesoscale wind speed, Um, representative for a 5×5km2 area block. The regional averaging of Um in the Ekman layer is done by geostrophic similarity methods, referring to a mesoscale block roughness obtained from maps through terrain classification. All wind data for the flat Netherlands are analysed, using a 30-year series of annual windiness indices for time-span adjustment of non-simultaneous data series. The evaluation leads to objectively interpolated seasonal surface wind maps, supplemented by wind distributions and extreme design wind speeds. Possible extension of the methodology to complex terrain is discussed.