Empirical angular distribution models (ADMs) are commonly used to convert satellite-measured radiances to top-of-atmosphere (TOA) radiative fluxes. This study compares two methods of developing ADMs: (1) the radiance pairs method (RPM), which composits ratios of near-simultaneous radiance measurements over the same scene to construct the ADMs; (2) the sorting-into-angular-bins (SAB) method, which estimates ADM anisotropic factors from the ratio of the mean radiance in each angular bin to the mean flux determined by direct integration of the mean radiances. Theoretical simulations and analyses of measurements from the CERES (Clouds and Earth's Radiant Energy System) satellite instrument show that the RPM method provides a better estimate of the true mean ADM for a population of scenes, while the SAB method is better suited for top-of-atmosphere flux estimation. The CERES results also show that a variable field of view size with viewing zenith angle can cause an ≈10% (relative) change in estimated all-sky mean albedo with viewing zenith angle. ¿ 1999 American Geophysical Union