A coal mine in South Africa had reached decant levels after mine flooding, where suspected mine water was discharging on the ground surface. Initial investigations had indicted a low-risk of decant, but when ash-backfilling was performed in the defunct underground mine, decant occurred. Ash-backfilling was immediately suspended as it was thought to have over-pressurised the system and caused decant. Contrariwise, a number of years later decant was still occurring even though ash-backfilling had been terminated. An investigation was launched to determine whether it was the ash-backfilling which had solely caused decant, or if additional contributing factors existed. Understanding the mine water decant is further complicated by the presence of underlying dolomites which when intersected during mining produced significant inflows into the underground mine workings. Furthermore, substantial subsidence has taken place over the underground mine area. These factors combined with the inherent difficulty of understanding unseen groundwater, produced a proverbial 1000-piece puzzle. Numerical groundwater modelling was a natural choice for evaluating the complex system of inter-related processes. A pre-mining model simulated the water table at the ground surface near the currently decanting area, suggesting this area was naturally susceptible for seepage conditions. The formation of a pathway from the mine to the ground surface combined with the natural susceptibility of the system may have resulted in the mine water decant. This hypothesis advocates that mine water was going to decant in this area, regardless of ash backfilling. The numerical groundwater flow model builds a case for this hypothesis from 1) the simulated upward flow in the pre-mining model and 2) the groundwater level is simulated above the surface near the currently decanting area. A mining model was then utilised to run four scenarios, investigating the flux from the dolomites, subsidence, ash-backfilling and a fault within the opencast mine. The ash-backfilling scenario model results led to the formation of the hypothesis that completing the ash-backfilling could potentially reduce the current decant volumes, which is seemingly counterintuitive. The numerical model suggested that the current ash-backfill areas reduce the groundwater velocity and could potentially reduce the decant volumes; in spite of its initial contribution to the mine water decant which is attributed to incorrect water abstraction methods. In conclusion, the application of numerical models to improve the understanding of complex systems is essential, because the result of interactions within a complex system are not intuitive and in many cases require mathematical simulation to be fully understood.