Monitoring of regional radiation budgets is limited by the availability of broadband measurement systems like the Earth Radiation Budget Experiment (ERBE) and the recent Clouds and the Earth's Radiant Energy System (CERES) and by the accuracy of the broadband fluxes derived from narrowband imagers on sun-synchronous and geostationary satellites. Incorporation of operational satellites would greatly reduce errors due to temporal sampling. Narrowband to broadband conversions that do not involve computer intensive radiative transfer models or those that do not require large apriori data set would be ideal. The narrowband to broadband models currently used in the Atmospheric Radiation Measurement (ARM) program geostationary products involve narrowband-to-broadband shortwave (SW) albedo conversions. These conversions are sensitive to the biases in the anisotropic models and are not dependent on cloud properties. The direct narrowband to broadband albedo conversion formulas include a solar zenith angle term but have a 10% relative RMS error. The considerable error in this approach results from its lack of terms accounting for cloud properties and vegetation changes and from inaccurate anisotropic models. The CERES Single Scanner Footprint TOA/Surface and Clouds (SSF) dataset provides coincident, collocated and co-angled SW and visible data at a nominal 10-km footprint. The dataset is limited to a 45° viewing zenith angle because of the Visible Infrared Scanner (VIRS) instrument and to data equatorward of 40° latitude because of the Tropical Rainfall Measuring Mission (TRMM) satellite orbit. This paper uses the CERES TRMM dataset and some preliminary CERES SSF data from the Terra satellite to derive improved narrow-to-broadband radiance models. The conversions are modeled as functions of angles, geotype, cloud optical depth and phase. These models can then be applied to pixel level imager satellites to compute more accurate high resolution broadband SW fluxes from VIS radiances.