Abstract: Many sensors are used to collect the comprehension information in landslides, which may reflect timevariance of a landslide. In the procedure of multisensor data fusion, some evaluation indexes are inconsistent. In order to eliminate this phenomenon, Bayes data fusion theory is employed. Analysis of landslides is carried out by using a decisionmaking fusion method. The results indicate that the landslide had experienced a slow deformation stage/ accelerated deformation stage/uniform motion deformation stage/ rapid deformation stage. It is consistent with the law of nature. The feasibility and effectiveness of the method had also been proved at the same time.