Soil erosion prediction models are of great significances in formulating reasonable strategies for soil and water conservation. Rill erosion is an important component of hillslope soil erosion. To well understand and predict rill erosion is essential for hillslope erosion prediction. In this study, a physical-process-based model for predicting rill erosion is developed, encompassing water flow continuity equation, hydrodynamics equation, sediment transport equation, for a well-defined rill channel. Finite element model was used to numerically solve the hydrodynamic and sediment continuity equations. Rational model parameters were used in this simulation study. A series of case studies were conducted with parameters determined of a silty loess soil from the Loess Plateau of China to compare outputs of the model with spatial changes in sedimentation processes for a better understanding of rill erosion dynamics. The comparisons of the experimental results with the simulation outputs of the model were used to validate the model. The model is capable of simulating the dynamics of rill erosion processes, with good prediction performance. Therefore, the model is proven to be valuable for further development of the dynamic prediction model for temporal and spatial variation of rill erosion.