This study aims to understand the effects of property variations on prediction discrepancies in the output capacity (i.e., output power) of deep coaxial borehole heat exchangers under different operational conditions. These properties include the specific heat and thermal conductivity of the water and the reservoir rock, as well as the density and dynamic viscosity of the fluid. Numerical simulation and grey relational analysis were combined to quantify the effects of the property variations on heat convection, heat conduction, and the heat exchanger’s predicted output capacity. The simulation results indicate that the average convective heat transfer coefficient of the water decreases, but the average thermal diffusivity of the rock increases with the extraction of geothermal energy. These changes further lead to an over-prediction of the output capacity. The grey relational analysis reveals that the flow rate and well depth are the principal factors causing over-prediction. The over-prediction increases with the rise of the injection flow rate and well depth. The results also show that the variability of the specific heat and the dynamic viscosity of the water are the two most significant factors causing a discrepancy in the predicted output capacity. The results from this research have highlighted the significance of considering property variations for a more practical prediction of the numerically modeled output capacity of deep coaxial borehole heat exchangers.