The thesis deals with issue aplication of artificial neural networks in geographic information system (GIS). The artificial neural network is represented as universal tool for creating a predictive models. This technique of making predictive models is not much extended in Slovakia, but in abroad it is relatively well known. In this thesis is artificial neural network applied to predict the risk of landslides in Trenčín and Žilina region. In pursuance of four different parameters (slope of relief, orientation of relief, land cover and subsoil geological structure) was created artificial neural network able to predict the risk of landslides. The product of this thesis is the model risk of landslides.