A radiance-to-flux conversion is needed to estimate radiative fluxes at the top of the atmosphere from directional measurements made by broadband (BB) radiometers on satellites. Such a conversion is known to be one of the major sources of error in the resulting instantaneous shortwave and longwave fluxes. This paper analyzes the possibility to improve the radiance-to-flux conversion for the longwave radiation when spectral information about the radiation is available through a set of narrow-band (NB) measurements. The study is based on a database of spectral radiance fields at the top of the atmosphere built using radiative transfer computation. The analysis of this database shows that there exists a certain degree of correlation between the angular and the spectral behaviors of the radiation field. According to the type and the accuracy of the spectral information, this correlation allows a 25–55% reduction of the error introduced by the radiance-to-flux conversion with respect to a simple model that uses onldy broadband information. The method discussed in this paper might be used when broadband radiometer and spectral imager data are available together like the combination of Geostationary Earth Radiation Budget (GERB) and Spinning Enhanced Visible and Infrared Radiometer Imager (SEVIRI) or the combination of CERES and MODIS.