Abstract Induced seismicity compromises the widespread deployment of geoenergy applications that contribute to mitigate climate change. In particular, the development of Enhanced Geothermal Systems (EGS) has been hindered by the risk of induced seismicity, mostly caused by hydraulic stimulation aimed at enhancing the permeability of deep hot crystalline rocks. Injection-induced seismicity has been traditionally attributed to fluid pressure buildup, which destabilizes fractures and faults. However, the largest seismic events commonly occur after the stop of injection, when pore pressure drops and both the magnitude and frequency of induced seismicity is expected to decrease. This counterintuitive phenomenon is not well understood. Yet, understanding the triggering mechanisms is the key to reliably forecast and manage induced seismicity. Here, we investigate the triggering mechanisms of co- and post-injection seismicity using coupled hydromechanical models, considering both a homogeneous and a fault-crossed domain, based on the case of Basel EGS (Switzerland). We find that the combination of pressure diffusion, poroelastic stressing, and static stress transfer explains the occurrence of induced seismicity, especially after the stop of injection, significantly better than the pore pressure alone. Considering a fault zone, which is more permeable and deformable than the surrounding rock, amplifies pressure diffusion along the fault and causes anisotropic variations of the stress field that lead to an increase in the seismicity rate that is orders of magnitude larger than for the homogeneous domain. These results point out that identifying the main geological structures through subsurface characterization is key to improve physics-based induced seismicity forecasting.