Abstract: Rain-induced landslide is a common geological hazard. The consequence of landslide hazard is closely related to its travel distance. The prediction of rain-induced landslide horizontal travel distance is the basis of quantitative risk assessment. This paper built a database through collecting 309 cases of rain-induced landslides systematically. According to the database, we statistically analyzed the horizontal travel distance, volume, vertical travel distance, initial slope gradient and aspect ratio. Then we developed a calibration method of rain-induced landslides travel distance empirical model considering both confined and unconfined landslides. The proposed method could not only get the prediction of horizontal travel distance, but also the model error and confidence interval. Single factor analysis shows that the correlation of vertical travel distance with horizontal travel distance is highest, following are volume, aspect ratio and initial slope gradient. Stepwise regression analysis shows that the prediction model of rain-induced landslides based on the vertical travel distance, volume and aspect ratio is the fittest.