On 27 December 2007 a M w 1.9 seismic event occurred in the Mponeng gold mine, South Africa. The seismic event's hypocentre, as well as the fore-and aftershocks were located with the JAGUARS acoustic emission array, placed at mining level 116. Thus, hypocentre depth (3509 m) and focal mechanism are very well constrained. Since no mining activity took place more than two days before the event, dynamic triggering due to blasting is ruled out. We investigate the hypothesis that static stress transfer due to excavation of the gold reef induced the event. To estimate these mining-induced stress changes, we set up a small scale (450 × 300 × 310 m 3 ) high resolution 3D geomechanical-numerical model, containing most of the seismic events detected. The model is made of the four different rock units present in the mine: quartzite (footwall), hard lava (hanging wall), conglomerate (gold reef), and diorite (dykes). The virgin in-situ stress, determined in nearby TauTona mine, is used to implement initial conditions. Elastic rock properties are taken from laboratory measurements. For the numerical solution, we are using the finite element method, with a discretised mesh of ∼1 million elements. From the computed 3D stress tensor we estimate the Coulomb failure stress change ( CFS) on the fault plane of the event in two ways: The first with shear stresses resolved in rake direction ( CFS slip ) and the second using the maximum shear stress ( CFS max ) on the plane. The CFS slip values are negative, indicating that, according to the model, mining activity brought slip in the resolved direction on the resolved fault plane further away from failure. Considering model uncertainties with respect to fault plane geometry, rake vector, and elastic parameters, this result remains stable. In contrary, the CFS max value is positive, indicating that, given the modelled stress is correct, a high potential for triggering a rupture is available.