The properties of the upper air winds used by the commercial aviation community are investigated. Analysis of the spectra shows that the forecasts have errors in both power and phase. However, the constant multiplier presently applied at The Met Office to the forecast winds before they are sent to the airlines remains best for remedying the speed underestimates characteristic of the forecast winds relative to those observed. We provide evidence for the success of this multiplier including the existence of a systematic negative speed bias as a function of the forecast wind speed. A constant multiplier greater than one (presently set at 1.04) always produces reductions in the traditional speed and vector wind rms scores, as well as an aviation specific rms relative flight time error. The investigations reveal that the constant multiplier is also optimal given the verification measures used; it must therefore remain integral to the operational processing of the forecast winds before transmission to the aviation community, at least for the foreseeable future.