The aim of this work was to rank the territory of the Caspian sandy natural plague focus (43) by the risk of epizooty emergence using the MaxEnt model.Materials and methods. The archival data on epizootic manifestations of plague over the past 35 years, aggregated by the Stavropol Anti-Plague Institute of the Rospotrebnadzor, the Dagestan, Elista, Astrakhan PCSs of the Rospotrebnadzor, were used for model design. 615 archive plague detection points were converted into the coordinate system (1980–2015). 87 publicly available bioclimatic variables BioClim were deployed to construct the MaxEnt model. Applied weather and climatic factors of the BioClim database are averaged over a multiyear period.Results and discussion. The MaxEnt model has a very high degree of reliability (AUC=0.975), with a sufficiently high predictive ability (AUC=0.973). According to the generated model, the Caspian sandy natural plague focus has a heterogeneous structure in terms of the probability of epizooty registration and can be divided into five zones. The most significant factors for the model are the following indicators: the average temperature of the wettest quarter, solar radiation in November, the average temperature of the driest quarter, the amount of precipitation in the coldest quarter, wind speed in May, the amount of precipitation in the wettest quarter, and the average air temperature in September. The data obtained allow for targeted search for plague epizootics and can be used to adjust boundaries of a surveyed natural focus in the future.