Two post-mortem human subjects were subjected to dynamic, non-injurious (up to 20 % chest deflection) anterior shoulder belt loading at 0.5 m/s and 0.9 m/s loading rates. The human surrogates were mounted to a stationary apparatus that supported the spine and shoulder in a configuration comparable to that achieved in a 48 km/h sled test at the time of maximum chest deformation. A hydraulically driven shoulder belt was used to load the anterior thorax which was instrumented with a load cell for measuring reaction force and uniaxial strain gages at the 4th and 8th ribs. In addition, the deformation of the chest was measured using a 16- camera Vicon 3D motion capture system. In order to investigate the chest deformation pattern and ribcage loading in greater detail, a human finite element (FE) model of the thorax was used to simulate the tests. The thorax FE model was positioned in a similar posture as used in testing and then the displacement time histories of belt ends were prescribed using the experimental data. The time histories of reaction force and the chest deformation predicted by the computational model were in good agreement with the force and deflection test data. Furthermore, chest deflection patterns predicted by experiments and FE simulations appeared to be sensitive to the geometry of the subject. Finally, reasonable correlation in terms of strain magnitude was observed between computational and experimental data for corresponding measurement locations. In addition to providing validation data for computational human models and dummies, the results of this study may lend insight into the development of advanced belt restraint systems.