An experimental study of pedicle screw fixation in human cadaveric vertebrae. The aim of this study was to experimentally characterize pedicle screw fixation under nonaxial loading and to analyze the effect of the surgeons' screw and placement choices on the fixation risk of failure. Pedicle screw fixation performance is traditionally characterized with axial pullout tests, which do not fully represent the various tridimensional loads sustained by the screws during correction maneuvers of severe spinal deformities. Previous studies have analyzed the biomechanics of nonaxial loads on pedicle screws, but their effects on the screw loosening mechanisms are still not well understood. A design of experiment (DOE) approach was used to evaluate 2 screw thread designs (single-lead and dual-lead threads), 2 insertion trajectories in the transverse and sagittal planes, and 2 loading directions (lateral and cranial). Pedicle screws were inserted in both pedicles of 12 cadaveric lumbar vertebrae for a total of 24 tests. Four sinewave loading cycles (0-400 N) were applied, orthogonally to the screw axis, at the screw head. The resulting forces, displacements, and rotations of the screws were recorded. In comparison to the other cycles, the first loading cycle revealed important permanent deformation of the bone (mean permanent displacement of the screw head of 0.79 mm), which gradually accumulated over the following cycles to 1.75 mm on average (plowing effect). The cranial loading direction caused significantly lower (P < 0.05) bone deformation than lateral loading. The dual-lead screw had a significantly higher (P < 0.05) initial stiffness than the single-lead thread screw. Nonaxial loads induce screw plowing that lead to bone compacting and subsequent screw loosening or even bone failure, thus reducing the pedicle screw fixation strength. Lateral loads induce greater bone deformation and risks of failure than cranial loads. N/A.