Surface bidirectional reflectance distribution function (BRDF) and albedo maps are derived from Moderate Resolution Imaging Spectroradiometer (MODIS) multiday surface reflectance composites with a 500-m spatial resolution (MODIS product MOD09A1/MYD09A1). The proposed method dubbed land cover--based fitting employs the processing of clear-sky reflectance data for similar land cover types and takes into account the magnitude of normalized difference vegetation index (NDVI). The BRDF is derived through the fitting of pixel data sorted into small bins according to the values of angular variables and NDVI. Robust statistical processing is applied to reduce the influence of noise and outliers. This method increases the success rate of the fitting process and enables more accurate monitoring of surface temporal changes during periods of rapid spring vegetation green up and autumn leaf fall, as well as changes due to agricultural practices and snow cover variations. The approach is specifically applied over the Atmospheric Radiation Measurement Program Southern Great Plains area. Results are compared to alternative BRDF/albedo products, such as the MOD43 albedo and Multiangle Imaging Spectroradiometer surface products that are derived through a pixel-based fitting process. A good agreement was generally found between different data sets. For example, the average biases in the visible and near-infrared bands are usually less than 0.01 and 0.02, respectively, and correlation coefficients are typically larger than 0.80. An analysis of these differences identifies some unique advantages of the proposed method, such as the ability to capture rapidly changing surface properties and an increased performance in the case of reduced number of clear-sky observations because of frequent cloudy conditions. Results suggest that the developed land cover--based methodology is valuable for the purpose of surface BRDF and albedo mapping using MODIS observations.