The seismic safety of slopes is generally estimated by a stability analysis using the conservative method such as the Fellenius method before slope failure. However, it is also important for the estimation of the seismic safety of the slopes to understand and simulate the characteristics of a landslide after slope failure. Therefore, a series of shaking table tests for the slope models with dry sand layer having lower strength are carried out, considering the effect of the slope gradient and width of the sand layer on the landslide after slope failure in order to estimate the seismic safety of the slopes. As a result, the behavior of the landslide masses depends on the type of slope models in the test i.e. the slope models exhibit the ductile failure, which can be simulated by a conservative analysis method such as Newmark method, and brittle one which is difficult for the conservative analysis methods to simulate. On the other hand, the simulation using the material point method, which is one of the particle in cell method and can deal with the large deformation of soils with clear constitutive laws, for the test is also carried out. Conventionally, the same trend of the behavior of landslide masses according to the type of slope models as the shaking table test results can be obtained by the material point method with the simple elasto-plastic Drucker Prager model as the constitutive law, considering the properties of the weak layer in the slope model.