Satellite remote sensing of clouds requires an accurate estimate of the clear-sky reflectance for a given scene to differentiate cloudy from clear pixels. The clear-sky reflectance can be observed at one set of solar zenith, viewing zenith, and relative azimuth angles for a limited set of conditions. To predict the clear-sky radiance at any other set of angles, it is necessary to know how the reflectance varies with the viewing and illumination angles. Typically, bidirectional reflectance distribution functions (BRDF) are used to predict the variation of normalized reflectance with viewing zenith and relative azimuth angles for a given solar zenith angle. Directional reflectance models (DRM) describe the variation of albedo with solar zenith angle. The BRDFs are used to predict reflectance from albedo. DRMs depend on many factors, especially surface type. The accuracy of narrowband DRMs has traditionally been limited by uncertainties in the BRDF and biases in the sampling patterns from geostationary and sun-synchronous satellites. The Clouds and Earth’s Radiant Energy System (CERES) project is using multispectral imagers, the Visible Infrared Scanner (VIRS) on the Tropical Rainfall Measuring Mission (TRMM) satellite and the Moderate Resolution Imaging Spectroradiometer (MODIS) on Terra, to provide cloud and clear-sky properties, in particular, a discrimination between clear and cloudy pixels. The TRMM, launched during late 1997, continues to provide coverage at all local hours between 37°N and 37°S over a period of 46 days, while the MODIS, operating since Spring 2000, has a 1030 LT Equatorial crossing time providing twice-per-day coverage in the Tropics and midlatitudes and higher temporal sampling in polar regions. A combination of the two datasets provides unprecedented angular sampling over all of the major surface types reducing the uncertainties associated with the BRDFs and the sampling from a combination of geostationary and sun-synchronous satellites. This paper presents the results of an analysis of the CERES clear-sky reflectances to derive a set of normalized top-of-atmosphere 0.65 and 1.6-µm DRMs for all major surface types using the combined MODIS and VIRS datasets. These models should be useful for many remote sensing and simulation studies.