The Beaufort equivalent scale of the World Meteorological Organization (WMO), used for decades to transform marine Beaufort estimates to surface wind speeds over the oceans, contains systematic errors that depend nonlinearly on the wind speed. Applying a revised scientific equivalent scale instead of the WMO scale produces significant changes in statistics of surface wind speed U over the ocean and, consequently, in all air-sea fluxes that are related to U. For the North Atlantic Ocean these biases are quantified as follows. The WMO scale underestimates climatological monthly means of U significantly: up to 1.6 m s−1in tropical latitudes throughout the year. In subpolar regions, differences are significant from spring through autumn and reach 1.3 m s−1. These regionally and seasonally different monthly biases are equivalent to an overestimate of the annual variation of U, which reaches 1.5 m s−1 in the westerlies. Local standard deviations may be overestimated up to 1.2 m s−1. The WMO scale underestimates climatological monthly estimates of latent heat flux up to 50 W m−2. (up to 25%). The bias of the mean annual North Atlantic evaporation rate is 0.3 m yr−1. The bias in annual net air-sea heat flux amounts to 27 W m−2, equivalent to an underestimate of the transequatorial oceanic heat transport by 1.15 PW (1 PW = 1015 W). Climatological monthly wind stress at the ocean surface is underestimated by more than 4.5 × 10−2 N m−2 (up to 50%) in the trade-wind region. Most existing regional and global air-sea flux compilations (including COADS) have been derived using the WMO scale. Hence, large biases are included in these compilations, although they can be partially hidden by an artificial increase of parameterization coefficients. The wind statistics revised according to a more accurate scale allow the application of bulk coefficients in accordance with newer experimental results from the open ocean. Therefore means and statistics of wind speed and climatological estimates of air-sea fluxes over the World Ocean need revision.