Abstract: A stochastic granule model is developed at mesomechanical level. A random distribution model for rockfill granule is generated by means of random simulation technique. In addition, the nonlinear contact algorithm is presented based on modified augmented Lagrangian algorithm. A simple linear softening damage model is employed to describe the stress and strain relationship of mesoscopic element, and the mechanical properties are distributed with a certain statistical function. The stochastic granule model of Maokou rockfill of Shuibuya rockfill dam is established. The simulated results show that the movement and deformation processes of rockfill granule agree well with the present experiments.