Satellite-based retrievals of sea-surface temperature (SST) hold great potential for augmenting and improving existing global climate analyses. However, optimal blending of satellite data with in situ measurements of SST requires an accurate estimate of the temperature difference between radiometric skin and bulk water sampled below the skin. Contemporaneous shipborne radiometer measurements and in situ measurements of SST were obtained during a cruise in the western Pacific and used to define the nighttime skin-bulk temperature difference. A strong trend of increasing temperature difference with decreasing wind speed was observed. Three published parameterizations of the skin effect have been tested against the ship observations, with excellent agreement achieved by one of these. Good results were also obtained when this scheme was forced by heat fluxes from the Met Office numerical weather prediction global analysis collocated with the observations in space and time. The implementation of a skin effect model in a global satellite data processing system was tested using collocated observations of SST from buoys (bulk) and along-track scanning radiometer 2 (skin) during 1995--1997. This is a vital step toward integrating accurate satellite retrievals of SST with traditional in situ data sources for climate variability studies.