Understanding the angular reflection properties of soils and quantifying their particle size distribution through physical models is valuable for ecological and agricultural applications in arid and semi-arid regions. An analytical radiative transfer (ART) model has considered the absorptivity, k, as a constant to simulate the angular reflection of soil at two measurement directions and invert the soil particle size. However, the limited viewing directions cannot reflect the key role of ART in characterizing the effects of particle size on angular reflection. In this study, an ART model was first assessed based on a multi-angular soil reflection dataset from six types of dry soil samples with different particle size. The absorptivity, k, an inverted model parameter, was found to be strongly dependent on the angular distribution of soil reflection for different soil types. Subsequently, we coupled the ART model with a semiempirical function that can characterize the angular distribution of k, to improve the model’s ability in simulating the multi-angular reflection of soil. The modified ART model (defined as MART-Soil) made it possible to generate satisfying simulated multi-angular soil reflections with different particle sizes from all six soil types. Moreover, the model parameters of MART-Soil have the potential for representing different soil types in the simulation. These results indicate that MART-Soil model can quantify angular reflection properties of soils with different particle size and may also be useful for characterizing the radiative transfer process in land surfaces.