The surface roughness of ice crystals is not routinely considered in current cloud retrieval algorithms based on pre-computed lookup libraries. In this study, the effect of ice crystal surface roughness on cloud effective size and optical thickness retrievals is examined by reflected solar radiation computations. The geometric optics method is employed to infer the single-scattering properties of ice crystals with both smooth and rough surfaces. The discrete ordinates radiative transfer (DISORT) model is used here to compute bi-directional reflectances for distributions of randomly oriented ice crystals with effective sizes varying from 20 to 130 μm. Cloud reflectance lookup libraries are generated at the typical visible channel wavelength of 0.66 μm and near infrared wavelength of 2.13 μm. The applicability of the lookup libraries relating to surface smooth and rough particles is demonstrated using a MODIS image located over the Western Pacific Ocean. An effective particle size and optical thickness is retrieved for every pixel containing clouds by using a combination of spectral channels that are absorbing (2.13 μm) and nonabsorbing (0.66 μm) for ice clouds.