The occurrence of landslides generally depends on complex interactions among a large number of partially interrelated
factors. It is appropriate to use multiple regression analysis for predicting landslides from a given set of independent
variables. The procedure of landslide hazard assessment by regression analysis, however, requires evaluation of the
spatially varying terrain conditions as well as spatial representation of the landslides. In this paper, the multiple
regression analysis was applied to predict landslides in Himi district from independent factors, such as geology,
slope-aspect, slope angle, land use and soil with Geographic Information System (GIS). Based on GIS, every factor was
classified into several clusters and then the statistical weight of every cluster was assigned for every factor respectively.
By the weights of five factors, the linear regression's coefficients of these input factors in landslide area were extracted
and assigned to the whole region, and then the susceptibility for the potential landslide was obtained to make the
landslide hazard assessment map. Geology and slope-aspect factors are the most important ones. Soil factor is not so
notable in this research region, though it may be significant in other regions. At last, the average susceptibilities map for
existing landslides was made for the engineers to do control work.