Purpose: The purpose of this experimental study was to compare fixation with hooks and screws inserted posteriorly. A digitalized analysis using finite element analysis was applied. Material and methods: We used seven human thoracic spines for this experimental study. We identified 49 pairs of two vertebrae. Traction was applied to rupture, the maximal force at rupture measured with an Instron. Fixations were made with four pedicle screws and two pediculolaminar clamps. For the digitalized study, the modellised vertebra was composed of 63000 nodes and 14000 elements. Calculations were made in the elastic domain using the finite elements abacus method. Results: Traction on the peidculolaminar clamp produced a fracture at the base of the pedicles in all cases. When screw fixation was used, there was a medial fissuration of the base of the pedicle. For hooks, pull-out force was 1108±510 Newtons. It was 820±418 Newtons for the 4-mm diameter screws and 1395±425 Newtons for the 5-mm screws. T5–T6 and T7–T8 assemblies ruptured more easily. The screw model demonstrated a concentration of the stress forces at the medial level of the pedicle, inside the spinal canal. Use of a long screw did not reduce stress significantly. The hook model demonstrated maximal stress force at the lower level of the pedicles. Discussion and conclusion: From a mechanical point of view, screw fixation is best, but this type of fixation did not fulfil all expectations. The results showed that the force for 4-mm screws is 23% weaker than for hooks and that 5-mm screws only provide a 12% better force than hooks. There are two mechanisms for pull-out, stripping of the bone threads, or rupture of the pedicles. The bone thread strips when the screw threads do not penetrate the cortical bone sufficiently because the screw is too small. On the contrary, larger screws risk injuring the pedicle. Pedicle rupture is observed for much higher stress force and constitutes the upper limit of resistance. This leads us to hypothesise that in most cases, screw pull-out occurs by bone thread stripping. Screws are less effective if they cannot be correctly anchored in the cortical, probably the cause of their relative weakness. The screw diameter should be chosen to adapt to the diameter of each pedicle. Stress forces would be transmitted better from the screw to the pedicle. The vertebrae are exposed to greater stress forces with hooks. The digitalised study confirmed that use of long screws crossing the entire vertebra did not provide a sufficient diminution of stress on the pedicles to justify their use