Predicting and controlling the subsurface seismic response to fluid injection, including CO (sub 2) storage, continues to be a challenge. Pre-existing, critically stressed faults in a crystalline basement rock that underlie injection intervals tend to be the most likely geologic features that may release seismic energy. However, the mechanisms of stress transfer from highly permeable storage units to very low permeability but fractured and faulted crystalline basement, where the origin of seismicity has been located, is not well understood. Here, we seek to provide a mechanistic explanation for the seismicity occurrence due to CO (sub 2) injection using a high-fidelity geological model that includes faults, crystalline basement rock, and overlying sedimentary rocks used for CO (sub 2) storage. We build a computational mesh that adapts to the surfaces of these geologic formations, and perform flow simulations to model the CO (sub 2) plume migration and reservoir pressures measured during the CO (sub 2) injection. The modeled reservoir pore pressures are then used for coupled flow and geomechanics modeling to assess their impact on the stability of the faults in our model. To identify faults that are more susceptible to releasing seismic energy, many variations of geologic descriptions are used, including pore pressure diffusion along intersecting fault segments and into the low-permeability basement. To validate the model predictions, a dataset from a CO (sub 2) storage demonstration project with measured and located microseismic events was used. The Illinois Basin Decatur Project injected 1 MtCO (sub 2) over 3 years with over 4700 located microseismic events (M<1.2). An intriguing observation is that the spatiotemporal evolution of the seismicity in Decatur is that some seismic events farther away from the injection well occur earlier than events located nearer to the well. To improve the match between observations and simulated results, we update the geological model iteratively to incrementally include fault planes determined from microseismic cluster analysis. In this manner, we aim at building a consistent geological model that explains the observed measurements and honors the available geological data to explain the observed seismicity occurrences.