Objective. The purpose of these studies is to determine, on the basis of the developed models describing seismic impacts, the most acceptable for acceptance from the point of view of the accuracy of the description of the seismic oscillation both in the shape of the envelope and in duration and frequency characteristics.Method. The use of empirical models in describing soil fluctuations as a random process has been considered in a number of papers. This paper presents the main simplifying assumptions used in this approach. The validity of empirical models and some difficulties encountered in modeling the oscillation process are investigated.Result. A methodology for assessing the reliability of the proposed impact models is proposed. The proposed method takes into account the soil - geological characteristics, as well as spectral data of instrumental recordings of known earthquakes.Conclusion. The seismic impact model, as well as the approximating modeling function, should be determined by the degree of responsibility of the building.