A well-known problem in hurricane forecasting and in the administration of hazardous weather warnings and advisories concerns what adjustment to make to flight-level reconnaissance wind observations in order to make them representative of sustained surface winds. To solve this problem, a study was initiated comparing NOAA reconnaissance flight-level winds to 10 m level observations from NOAA's oceanic buoys and platforms. A data base was created that consisted of comparisons made whenever the aircraft observation was: (1) within 10 km radial seperation from the surface platform (in a storm-relative coordinate system), (2) within ±4 h of the surface observation time, (3) within ±2 h of the surface observation time. The data base contains all storms flown by NOAA aircraft in the vicinity of the Atlantic and Gulf of Mexico buoy network over the eleven year period from 1975–1986. Comparisons from these criteria are discussed in terms of the ratio of the buoy-measured wind speed (VB) to the aircraft measured wind speed VA. Results indicate that the 10 m level surface winds over water were within 55–85% of the winds measured by the reconnaissance aircraft. The ratio VB/VA depended strongly on the low-level atmospheric stability as indicated by the buoy air-sea temperature difference.