Previously published estimates of the surface turbulent fluxes over the North Atlantic Ocean have been compared by applying the calculation methods used by each author to a common dataset of ship observations. The major differences between the various flux estimations were due to whether or not the data were corrected for observation height, the method of calculating averaged fluxes and the choice of transfer coefficient. By correcting the ship observations obtained from the Voluntary observing ship Special Observing Programme–North Atlantic for known observational biases an estimate of the correct values for the fluxes has been made. The previous studies were found to have overestimated the North Atlantic sea to air heat transfer by up to 30%. All of the schemes overestimated the annual cycle, giving fluxes approximately equal to the best estimate values in the summer but much higher fluxes in the winter. The use of transfer coefficient values much larger than the values determined by air-sea interaction experiments could not be justified by either the effects of measurement errors or by fair weather bias. The lower turbulent flux values are compatible with a balanced North Atlantic heat budget, given the uncertainty in the other flux terms.