Mechanical constraints to spine motion can arise in a variety of real-world situations such as when shoulder belts prevent anterior translation of the thorax during automotive collisions. The effect of such constraint on spinal column–spinal cord interaction during injury remains unknown. The purpose of the present study was to compare maximal dynamic spinal canal occlusion, measured via a specialized transducer, in cadaveric upper thoracic spine specimens under a variety of anterior–posterior constraint conditions. Four injury models were produced using 24 cadaveric spine specimens (T1–T4). Incremental compressive trauma was applied under constrained (i.e. blocked anterior–posterior translation) flexion-compression, pure-compression and extension-compression, and under unconstrained (i.e. free anterior–posterior translation) flexion-compression. All displacements were applied at 500 mm/s. For all three constrained trauma groups, complete transducer occlusion occurred between 20 and 30 mm of compressive displacement. The extension–compression caused transducer occlusion significantly less than the other constrained models (p<0.022) at 20 mm compression. For unconstrained flexion-compression, a compression of up to 50 mm resulted in a mean of 26% transducer occlusion. The constrained pure-compression tests led to burst fracture with significant body height loss at T2. The constrained flexion-compression and extension-compression tests caused fracture-dislocation injury at the T2–T3 level. Constrained trauma clearly led to more spinal canal occlusion than the unconstrained in these models, and more severe injury to the spinal column. The results add to our understanding of the effect of column injury pattern on spinal cord injury. This information has clear implications for the design of injury prevention devices.