Abstract: Intraday and cumulative rainfalls were chosen as the meteorological factors of the prediction model through comparing the relationship between the landslides and the precipitation of the early days. The attenuation coefficient of cumulative rainfall was 0.9. The binary logistic regression prediction method is established based on 120 rainfall-induced-landslides data in the Yucheng district, Yaan city in Sichuang province. Analysis is done with the independent variables of the intraday and cumulative rainfall and probability prediction equation is achieved. On this basis, a critical expression of rainfall-induced-landslides is obtained through the scatter plot. The test samples of 5 rainfall-induced-landslides and 5 non-landslides reach up 100 percent accuracy. The predictive values of the ten tests samples are in line with requirements. The prediction of the rainfall-induced-landslides can be provided by forecast equation and critical expression.