Nowadays the evaluation of physical vulnerability of the built-up interacting with landslides is a topic of great interest, due to the increasing number of disasters. At the state of art different approaches are spread for assessing physical vulnerability. Prominent among these are methods based on the damage data collected in the post-event stage, since they provide a clear understanding of the damage mechanisms, as well as of the building behaviour interacting with the landslide. In this study a statistical methodology was developed for estimating the physical vulnerability of buildings exposed to slow-moving landslides. The method was applied on a dataset of 170 buildings. Data collection was based on post-landslide events field surveys carried out in five municipalities of the Campania region, in Southern Italy. After the detection of the most influential vulnerability parameters, a multinomial logistic regression analysis was applied on the database for establishing a damage predictive model. In the end the model has been validated on the dataset in order to evaluate the response.