The present study aims to numerically estimate the coefficient of kinetic friction for fault surfaces subjected to confining stress ranging from 20 MPa to 100 MPa, assuming depths from 1000 m to 5000 m, in order to obtain a better understanding of the dynamic behaviour of faults during seismic events. To achieve this, a 3D numerical model including an inclined fault was constructed, where seismic events were simulated under dynamic conditions in a circular region on the fault by employing Barton's shear strength model and a slip-weakening law to consider both the static friction as well as dynamic slip-weakening behaviour. Subsequently, five cases were analyzed whilst varying the in-situ stress state applied to the model. For each case, fault-slip was simulated, and seismic efficiency was computed from the slip rate during the event, fracture energy, and frictional energy loss of the fault. The coefficient of kinetic friction was then calibrated so that the seismic efficiency of the simulated seismic events becomes 0.06, which constrains the kinetic friction in the calibration. Finally, the relation between the calibrated kinetic friction coefficient and the maximum slip rate at the center of the hypocenter was examined, and a fitted line was derived based on a regression analysis. The fitted line was compared to previous experimental results as well as those derived from a rate- and state dependent law. The comparison demonstrated that for high slip rates above 2.5 m/s, the fitted line gives the coefficient of kinetic friction more comparable to the experiment than the rate and state dependent law, while for low slip rates of the order of pm/s, the rate- and state-dependent law gives a more accurate estimation of kinetic friction, i.e. the proposed line can underestimate the change from static to kinetic friction. This suggests that seismic efficiency can be used to estimate the coefficient of kinetic friction for rock discontinuities that undergo a slip at the rate of more than 2.5 m/s.