Injection of CO2 into deep hydrothermal reservoirs is considered as an attractive way for heat extraction, which has been proved in the previous studies. However, how different geological factors govern the heat extraction efficiency by CO2 circulation are not clearly understood. In this study, the mechanisms of geological factors in controlling the heat mining process by CO2 injection are numerically explored. The effects of horizontal correlation length, degree of permeability heterogeneity, permeability anisotropy, and the magnitude of mean permeability on the heat extraction performance are systematically investigated. The simulation results show that heat extraction using CO2 as working fluid can obtain a high heat mining rate, but the CO2 plume in the subsurface exhibits severe segregation effect. Heat extraction by CO2 recycling is more sensitive to the variation of horizontal correlation length in comparison with that by water circulation. The increase of horizontal correlation length and mean permeability can effectively promote the heat mining rate. The increase of permeability heterogeneity and anisotropy can cause a reversed trend of heat mining rate variation. The findings from this study may provide important implications for the selection of feasible hydrothermal reservoirs or deep high water cut oil reservoirs for heat extraction by CO2 recycling.