The article presents a new approach for identifying previously unexplored archaeological sites of the Scythian period (7th-3rd centuries BC) in the territory of Southwestern Tuva through the application of modern research methods, namely neural networks. To achieve this objective, a database was created comprising archaeological sites of the Scythian era located within the study area. The database was compiled both from literature sources and from the authors' own field research. The analysis involved 639 Scythian-period sites. Based on the SRTM digital elevation model in ArcGIS 10.2, using the Spatial Analyst extension, as well as in the Python 3.7 environment, a geoinformation analysis was carried out to assess the distribution of archaeological sites across nine landscape parameters: absolute elevation, slope gradient, slope aspect, proximity to watercourses, elevation above the nearest watercourse, solar radiation intensity in December and June, visibility of mountain peaks, and distance from mountain peaks. The choice of parameters was informed by patterns of archaeological site distribution observed in the landscapes of Southwestern Tuva and Southeastern Altai during field expeditions. Subsequently, variance analysis was performed to determine the significance of each parameter. Based on the obtained data, a feedforward neural network was trained, and a predictive model of potential locations of previously unexplored archaeological sites was constructed. Verification of the predictive model was conducted using satellite imagery provided by the World Imagery resource within ArcGIS Online. Approximately 150 predicted areas, each measuring 250×250 meters, were visually inspected. In 48 cases, previously unknown archaeological objects (kurgans) were discovered. In conclusion, considering all the acquired data, a cartographic scheme of the settlement area of populations during the Scythian period was developed.