Abstract:The groundwater system in landslides is a complicated one influenced by many factors. According to the relation between groundwater level and the factors that affect the variation of groundwater level in landslides,the first purpose of groundwater predition is to find out the reasonable boundary condition that is basis for stability analysis of landslides. The second purpose of predition is to find out the variation of stability coefficient versus groundwater level. For these purposes,artificial neural network model which adopts radial basis function is developed,based on the relationship between groundwater level and its influential factors used for the predition of groundwater regime.Case study indicates that the precision of the developed model is rather high and its popularization significance is better than the others models.