The “Best Track” committtee of the Tropical Prediction Center recently reclassifed Hurricane Andrew as a Category Five storm at landfall in South Florida, based on an interpretation (that peak surface winds are ~ 90% of the flight-level winds at the 700 mb level) of open ocean GPS sonde measurements made since 1997. The accuracy of applying a single flight-level wind reduction factor is assessed by describing the distribution of reduction factors. Reducing eyewall aircraft measurements to the surface with the 90% method overestimates key eyewall surface observations at Fowey Rocks and Perrine. One possible reason is that open ocean marine roughness is much smoother than that on the coast. Very few sondes are available over shallow waters but coastal tower observations suggest that coastal marine roughness differs greatly from open ocean conditions. In addition, recent measurements from the GPS sondes and Stepped Frequency Microwave Radiometer (SFMR) suggest that in individual storms, the reduction factor may follow an asymmetric pattern that is influenced by relative flow patterns associated with a storm passing through a sheared environment. This pattern is consistent with recent modeling results of Kepert, which predict (relative to the storm direction) a maximum reduction factor on the left and minimum on the right. Another possible reason a fixed reduction factor fails in Andrew is related to the concentic eyewall process. There is ample evidence that Andrew was similar to historical storms (e.g. Hurricanes Inez in 1966, Allen in 1980, Gloria in 1985, Gilbert in 1988) that differ from those in the GPS sonde database. Intense hurricanes undergoing a contracting eyewall process have near-vertical eyewalls with maximum winds near uniform from the top of the boundary layer to the middle atmosphere. These data provide insight on the complexity of hurricane wind fields at landfall and suggest that a “one size fits all” reduction factor may not fit storms like Andrew.