Abstract: Aiming at the problem of spatial variability of soil parameters, a numerical method, Monte Carlo material point method, was proposed to quantitatively assess the risk of the whole process of landslide. Based on four quantitative evaluation parameters, including sliding time, sliding kinetic energy, sliding mass ratio and sliding distance, the whole process risk of landslide was quantitatively assessed under different probability distributions of soil cohesion and internal friction angle. The results show that the whole process risk of landslide can be more accurately assessed by considering soil variability when compared with the deterministic analysis. The simulation results under normal distribution and lognormal distributionare are close with eath other, but different from those under triangular distribution and uniform distribution. The Monte Carlo material point method can effectively carry out the quantitative risk assessment of the whole process of landslide when the cohesion satisfies a log-normal distribution and the internal friction angle is of normal distribution.