Landslide susceptibility mapping is very crucial for planning and development in a disaster prone region in Nepal. Nepal is one of the landslide prone countries of the World. Very high relief, steep slopes, complex geology and diverse vegetation cover has made Nepal vulnerable to landslides. Some national level and individual research initiatives have been published about landslide process, mechanisms and hazard zonation. But there are a few studies carried out in the Western region of Nepal which is one of the landslide affected regions of the country. The main purpose of the study was to prepare landslide susceptibility maps of the five sample sites of Western Nepal (Palpa/Gulmi, Palpa, Baglung/Myagdi, Parbat and Kaski) and validate the result by model replication. A number of qualitative and quantita-tive landslide susceptibility assessment methods exist to evaluate the landslide susceptibil-ity. They are briefly reviewed here. In this study, a bivariate statistical method-Landslide Nominal Susceptibility Factor (LNSF) is employed to analyze the data. Database of related landslide casual factor maps: slope, lithology and land cover were derived in Arc GIS envi-ronment. Landslide Susceptibility Index (LSI) maps were generated by establishing a rela-tionship between landslides and the factor maps. Distinct color variations between bounda-ries of the lithology were visible in the maps. The reason for this was the weighting process. The analysis of relationship between landslide inventory and the thematic maps in-ferred that casual factors such as slope gradients, lithology and land use pattern contributed to landslide process. But the contribution of each parameter was site dependent. The opti-mum range of slope gradient from where the landslide distribution recessed varied between study areas. The landslides in sedimentary rocks failed in lower angles. Terai, Siwaliks, Lower Nuwakot, Upper Nuwakot and Tansen lithological zones composed of alluvial soils, sedimentary and metamorphic rock; were prone to landslides. Although Higher Himalayan zone consists of relative stable rock masses; it consisted of some landslides. High slope gradients and bare rocks induce landslide in this zone. The intensive land use in fragile geology also contributed positively for landslide failure in lower slope angles. In Pal-pa/Gulmi and Palpa, Parbat and Kaski cropland and forests were most sensitive to land-slides. Large proportions of landslides were confined to cropland and forests. The forests in these areas are intensively used as a result forests are degraded. The other most im-portant aspect is that presence of forests in highly prone lithology might induce landslides due to added canopy and stem weights of trees during monsoons. Another reason is that the vegetated areas in steep slopes with little soil also affects the soil-root anchorage of the vegetation and consequently affect the landslide process. The effect of each parameter was analyzed using rating curves. The analyses indi-cated that the effect of susceptibility parameters is site specific. The effect of lithology was distinct in Palpa/Gulmi, Palpa and Baglung/Myagdi while slope, lithology and land cover had similar effect for Parbat and Kaski. Including an input parameter in susceptibility anal-ysis does change the output. The exclusion of land cover when susceptibility mapping did not show any changes in the rating curves. Therefore, identifying the most influential pa-rameter is important for susceptibility modeling.Model replication proved moderately successful for areas of similar lithology con-ditions. The rating curves were slightly higher than the hypothetical diagonal curve of 0 to1. The first 20% high susceptibility zone occupied around 38% and 39% of the landslides in Palpa/Gulmi and Parbat respectively. Model validation produced moderately satisfactory results when area distribution by susceptibility class between calibrated and validated model was compared. An increase in percentage of area in very low and low susceptibility classes in both replications was observed. In Palpa/Gulmi, calibrated map occupied 33% of the total area while model rep-licated map yielded 37% of the total area. In Parbat, calibrated map hosted 6% of the total whereas model replicated map owned 36% of the total area. Larger areas in lower suscep-tibility zones have implications over land use management. Larger area in lower suscepti-bility zones means large proportion of area can be used. Similarly, it improved the predic-tive capacity of the model by reducing the area of most susceptible zones by 32% com-pared to the area predicted in the original map (susceptibility map developed using the landslides of the same area) in Parbat.