When cloud properties are retrieved from satellite observations, current calculations apply one-dimensional (1D) theory to the three-dimensional (3D) world: they consider only vertical processes and ignore horizontal interactions. This paper proposes a novel approach that estimates 3D effects in cloud optical thickness retrievals. The proposed method combines visible and thermal infrared images to see whether 3D radiative effects make clouds appear asymmetric—that is, whether cloud surfaces tilted toward the sun are systematically brighter than surfaces tilted away from it. The observed asymmetries are then used to estimate 3D effects for 1-km-size pixels as well as 50-km-size areas. Initial results obtained for Moderate-Resolution Imaging Spectroradiometer (MODIS) images reveal that 3D effects cause abundant uncertainties in the 1-km-resolution 1D retrievals. Averaging over 50 km by 50 km areas greatly reduces the errors but does not remove them completely. Conservative estimates show that the mean optical thickness values are biased by more than 10% in 10% of the areas, and the errors in the areas’ standard deviation values are more than 10% in about 20% of areas.