In the early 1960's, N.G.W. Cook, using an underground network of geophones, demonstrated that most Witwatersrand tremors are closely associated with deep level gold mining operations. He also showed that the energy released by the closure of the tabular stopes at depths of the order of 2 km was more than sufficient to account for the mining-induced earthquakes. I report here updated versions of these two results based on more modern underground data acquired in the Witwatersrand gold fields. Firstly, an extensive suite of in situ stress data indicate that the ambient state of crustal stress here is close to the failure state in the absence of mining even though the tectonic setting is thoroughly stable. Mining initially stabilizes the rock mass by reducing the pore fluid pressure from its initial hydrostatic state to nearly zero. The extensive mine excavations, as Cook showed, concentrate the deviatoric stresses, in localized regions of the abutments, back into a failure state resulting in seismicity. Secondly, there appears to be two distinct types of mining-induced earthquakes: the first is strongly coupled to the mining and involves shear failure plus a coseismic volume reduction; the second type is not evidently coupled to any particular mine face, shows purely deviatoric failure, and is presumably caused by more regional changes in the state of stress due to mining. Thirdly, energy budgets for mining induced earthquakes of both types indicate that, of the available released energy, only a few per cent is radiated by the seismic waves with the majority being consumed in overcoming fault friction. Published by Elsevier Science Ltd.In the early 1960's, N.G.W. Cook, using an underground network of geophones, demonstrated that most Witwatersrand tremors are closely associated with deep level gold mining operations. He also showed that the energy released by the closure of the tabular stopes at depths of the order of 2 km was more than sufficient to account for the mining-induced earthquakes. I report here updated versions of these two results based on more modern underground data acquired in the Witwatersrand gold fields. Firstly, an extensive suite of in situ stress data indicate that the ambient state of crustal stress here is close to the failure state in the absence of mining even though the tectonic setting is thoroughly stable. Mining initially stabilizes the rock mass by reducing the pore fluid pressure from its initial hydrostatic state to nearly zero. The extensive mine excavations, as Cook showed, concentrate the deviatoric stresses, in localized regions of the abutments, back into a failure state resulting in seismicity. Secondly, there appears to be two distinct types of mining-induced earthquakes: the first is strongly coupled to the mining and involves shear failure plus a coseismic volume reduction; the second type is not evidently coupled to any particular mine face, shows purely deviatoric failure, and is presumably caused by more regional changes in the state of stress due to mining. Thirdly, energy budgets for mining induced earthquakes of both types indicate that, of the available released energy, only a few per cent is radiated by the seismic waves with the majority being consumed in overcoming fault friction.