Commonly, experimental data on surface directional reflectance are not properly corrected for atmospheric effects. In this paper, we describe a new method for rigorous retrieval of the bidirectional reflectance distribution function (BRDF) from ground-based measurements. It obtains the pointwise BRDF in the view directions and the best fit parameters of the approximating analytical model. Unlike previously developed algorithms, it can perform a retrieval using directional observations made at a single solar angle. This improvement allows one to process experiments with incomplete sets of measurements, such as those contaminated by clouds. The paper describes the performance of the method and its sensitivity to atmospheric parameters. Using rigorous direct modeling of the reflected radiance and developed inversion algorithm, we performed an analysis of BRDF errors, typical for measurements. It shows that the BRDF shape distortions due to diffuse irradiance are considerable in the visible and near-IR, even for low aerosol content in the atmosphere. This result confirms that an accurate atmospheric correction should be a required processing step in the experimental research of anisotropic surface reflectance.