Computational models of anthropomorphic test devices (ATDs) can be used in crash simulations to quantify the injury risks to occupants in both a cost-effective and time-sensitive manner. The purpose of this study was to validate the performance of a 50th percentile THOR finite element (FE) model against a physical THOR ATD in 11 unique loading scenarios. Physical tests used for validation were performed on a Horizontal Impact Accelerator (HIA) where the peak sled acceleration ranged from 8-20 G and the time to peak acceleration ranged from 40-110 ms. The directions of sled acceleration relative to the THOR model consisted of -GX (frontal impact), +GY (left-sided lateral impact), and +GZ (downward vertical impact) orientations. Simulation responses were compared to physical tests using the CORrelation and Analysis (CORA) method. Using a weighted method, the average response and standard error by direction was +GY (0.83±0.03), -GX (0.80±0.01), and +GZ (0.76±0.03). Qualitative and quantitative results demonstrated the FE model's kinetics and kinematics were sufficiently validated against its counterpart physical model in the tested loading directions.