Anthropomorphic test devices (ATDs) should accurately depict head kinematics in crash tests, and thoracic spine properties have been demonstrated to affect those kinematics. To investigate the relationships between thoracic spine system dynamics and upper thoracic kinematics in crash-level scenarios, three adult post-mortem human subjects (PMHS) were tested in both Isolated Segment Manipulation (ISM) and sled configurations. In frontal sled tests, the T6-T8 vertebrae of the PMHS were coupled through a novel fixation technique to a rigid seat to directly measure thoracic spine loading. Mid-thoracic spine and belt loads along with head, spine, and pectoral girdle (PG) displacements were measured in 12 sled tests conducted with the three PMHS (3-pt lap-shoulder belted/unbelted at velocities from 3.8 - 7.0 m/s applied directly through T6-T8). The sled pulse, ISM-derived characteristic properties of that PMHS, and externally applied forces due to head-neck inertia and shoulder belt constraint were used to predict kinematic time histories of the T1-T6 spine segment. The experimental impulse applied to the upper thorax was normalized to be consistent with a T6 force/sled acceleration sinusoidal profile, and the result was an improvement in the prediction of T3 X-axis displacements with ISM properties. Differences between experimental and model-predicted displacement-time history increases were quantified with respect to speed. These discrepancies were attributed to the lack of rotational inertia of the head-neck late in the event as well as restricted kyphosis and viscoelasticity of spine constitutive structures through costovertebral interactions and mid-spine fixation. The results indicate that system dynamic properties from sub-injurious ISM testing could be useful for characterizing forward trajectories of the upper thoracic spine in higher energy crash simulations, leading to improved biofidelity for both ATDs and finite element models.