Thirteen fresh cadaver spinal segments (fifth thoracic to fifth lumbar vertebra) with a posterior ligamentous defect at the twelfth thoracic-first lumbar level were tested to failure in flexion, creating an unstable fracture at the twelfth thoracic and first lumbar vertebrae. The fractured spines were then instrumented with Harrington distraction rods and retested. Spines instrumented at the eleventh thoracic to second lumbar vertebrae were not significantly stronger than uninstrumented specimens (with a posterior ligamentous defect). Repositioning the upper hook from the eleventh to the tenth thoracic vertebra increased the failure moment by an average of 36 per cent. Instrumentation of the spine at either level protected the spinal column against distraction at the fracture site until a threshold movement was reached. Moving the upper hook from the eleventh up to the tenth thoracic vertebra increased the threshold moment by an average of 65 per cent. Our recommended hook placement at the tenth thoracic and second lumbar vertebrae tends to reduce tilting of the upper vertebra, changing the mode of failure from slip-out of the upper hook (with partial laminar fracture) to total laminar fracture. CLINICAL RELEVANCE: The presence of internal fixation rods in itself does not make the unstable fractured spine stronger in a flexion mode. The level of hook placement above and below the fracture is important in determining with bending strength of the instrumented spine. When the use of Harrington distraction rods is contemplated for internally stabilizing a fractured spine, a specific level of hook placement, three laminae above and two laminae below the point of instability, is recommended. This configuration allows the instrumentation to maintain increased spinal stability in a flexion mode by reducing the tendency for the upper hooks to "back out" and lose their hold secondary to vertebral tilting.