In estimating parameters of non-linear models for landslide prediction,the traditional method is to develop into Taylor series at the parameter approximate value and only obtain one item,and then handle by linear model.Because of ignoring the items of two steps and above two steps in linearization,model errorwiu occur inevitably.By introducing Gauss—Newton method and taking Saleshan landslide for an example,Gauss—Newton method is used to optimize parameters of the two non-linear models on the basis of establishing grey GM(1,1) model and Verhulst model of the landslide.Results indicate that the prediction precision of the two models after optimizing parameters is obviously higher than that of the models before optimizing parameters.So,it is an effective and feasible method for improving landslide prediction precision to use Gauss—Newton method to optimize parameters of non-linear models.