Sea surface temperature data from a shipborne radiometer were used to assess the thermal skin effect parameterization schemes of Saunders <1967>, Hasse <1971>, Schl¿ssel et al. <1990>, and Soloviev and Schl¿ssel <1994>. Under low-wind, high-insolation conditions, the presence of a near surface thermocline resulted in an apparent skin effect which depended on the history of the surface heating. Neglecting these cases, the Saunders <1967> and other schemes which omit solar radiation were the most effective. The observed variation with wind speed of the adjustable coefficient, λs, in the Saunders formulae was similar to that predicted by Soloviev and Schl¿ssel <1994>, but the values were significantly higher, resulting in a greater magnitude of the observed skin effect. For the conditions occurring in this data set the better formulae predicted the skin effect with a standard deviation of ¿0.16 ¿C compared to a variation of the observed skin effect of nearly ¿0.3 ¿C. ¿ American Geophysical Union 1996