Susceptibility assessment of landslides over a large area depends on the basic spatial unit of mapping, usually by using grid cell or slope unit. Both units are used in this study for the assessment of small shallow and clustered landslides in vegetated slopes in Malipo, southwest China. Information value (IV) model was used to generate landslide susceptibility assessment map, while improved information value (IIV) model was used to determine whether the mapping unit is at risk of landslide. Seven factors, including slope angle, slope aspect, elevation, normalized difference vegetation Index (NDVI), Soil Moisture Content (SMC), distance to river and road were used as landslide influence factors. The Area under curve (AUC) values of the slope unit IIV, IV and grid cell were 0.814, 0.802 and 0.702 respectively for success rate. For prediction rate, the AUC values of the slope unit and grid cell were 0.803(IIV), 0.790(IV) and 0.699 respectively. Our results showed slope unit is more suitable than grid cell for assessing susceptibility of Small, Shallow and Cluster Landslide. Improved information value model increases the accuracy of susceptibility assessment model for this characteristic landslide.