ABSTRACT:A new analysis of the neutral heat transfer coefficient C HN on data fromÖstergarnsholm is presented, which is primarily based on a limited set of measurements with the very accurate MIUU (Meteorological Institute of the University of Uppsala) instrument, but with additional information from an extensive set of measurements with Solent sonic R2. Sonic data are, however, used with great caution, since for wind speed U above 10 m s −1 , a strongly windspeed-dependent correction is shown to be required. This error is roughly proportional to (U − 10) for sea-air temperature differences less than 4-5 K. For a larger temperature difference, no correction appears to be necessary in the wind speed range 10-15 m s −1 .We infer from our data that for conditions when unstable and near-neutral conditions prevail, measurements of the sea surface -air temperature difference are accurate to within 0.1 K at our site. This means that data for a range of relatively small temperature differences (0.5-1.5 K) which were often rejected in previous studies could be retained. It is observed that a rapid increase of C H and C HN occurs in that range.For wind speed above 10 m s −1 , C HN is observed to increase rapidly with U 10 . During those conditions, the wave field at the site is known to have characteristics very similar to those in deep-sea conditions. In a previous analysis of data fromÖstergarnsholm, it was speculated that observed high C HN values could be due to spray. Calculations with a spray model showed, however, conclusively that for wind speeds less than 14 m s −1 , the spray effect on the sensible heat flux is expected to be small. The high C HN values must instead be due to dynamic effects.It is demonstrated that when the Obukhov length L is less than about −150 m a regime with very specific characteristics ensues. This regime is dominated by surface-layer scale eddies, which cause Monin-Obukhov relations for the exchange of sensible heat to break down. The characteristics of this surface-layer regime are treated in detail in the companion paper.The rise of C HN with wind speed is shown to be closely related to a corresponding increase of z 0T with roughness Reynolds number for winds above 10 m s −1 . This means that during those conditions, traditional surface renewal theory for heat is no longer valid. It is suggested that this, in turn, is a result of increasing importance of wave-breaking with increasing wind and with a possible link to processes in near-surface atmospheric layers in the regime with −L > 150 m.