A validation experiment and resulting potential improvements to the operational satellite aerosol optical thickness product at the National Oceanic and Atmospheric Administration/National Environmental Satellite Data and Information Service (NOAA/NESDIS) are presented. An earlier paper described a set of Sun photometer measurements collected from the Societ R/V Akademik Vernadsky during its cruise in the Atlantic Ocean and Mediterranean Sea from September to December 1989. The accuracy of the Sun photometer aerosol optical thickness was proven acceptable for use as a ground truth standard for validation of the NOAA product. This paper describes the validation methodology and the results of its application to the NOAA 11 satellite product. A systematic underestimation in the operational values by about 35%, relative to the ship truth, is found. Causes for this discrepancy are examined, emphasizing the importance of careful satellite instrument calibration, and a revision of the oceanic reflectance model used in the retrieval algorithm. It is shown that the remaining systematic underestimate in satellite aerosol optical thickness can be attributed only to the aerosol model used in the retrieval. Additional checks of this conclusion using independent data sets are underway. If confirmed, a fundamental revision of the presently used aerosol model would be required. An example of a simple adjustment to the present aerosol model which successfully removes the bias is given, based on the assumption of an absorbing aerosol. ¿ American Geophysical Union 1995