A patient-specific poroelastic model of the brain was constructed in COMSOL Multiphysics and evaluated for its usability in a clinical setting. Image processing of magnetic resonance (MR) images of a standard (uninjured) brain and a computed tomography (CT) scan of a patient with a subdural hematoma was used to make an estimation of the shape patient’s pre-injury brain and obtain a deformation map describing the displacement due to the hematoma. A finite-element mesh of the normal brain was generated from a standard MRI-based brain atlas. The steady-state solution of the normal brain was similar to that found in a previous study. The steady-state solution of the deformed brain had the same pressure distribution as the normal brain, which was predicted by analysis of the equations. When the hematoma was simulated over 15000 seconds, the progression of the pressure over time seemed qualitatively plausible. The strain in the steady-state deformed brain and time-based studies generally behaved as expected, but a few regions near the surface of the brain adjacent to the hematoma had unexpectedly high values. When the magnitude of hematoma deformation was altered, it was seen that intracranial pressure, maximum pressure and strain had an exponential-recovery relationship over time for all factors. Strain rate decreased exponentially with time. The intracranial pressure, maximum pressure and strain had a linear relationship with a scale factor that was used to change to magnitude of deformation of the outer surface. Cortical cell death increased exponentially with scale factor and with time. Hematoma volume had a linear relationship with scale factor. Simulated pressure-volume curves did not have the same shape as experimental curves and the observed deformation had noticeable differences in mid-line shift when compared with the real brain. Conditions of hyper- and hypotension had big effects on the pressure, but not the strain. The results indicate that the model has the potential to be a good tool in brain injury evaluation, but more work must be done to increase accuracy and to validate the model.