Flexibility tests and finite element analyses were performed for the biomechanical evaluation of diagonal transfixation in pedicle screw instrumentation. To assess the biomechanical advantages of diagonal transfixation compared with conventional horizontal transfixation. A few pedicle screw instrumentation systems allow the use of cross-links in the diagonal direction. Such a diagonal transfixation is anticipated to improve the surgical construct stability, but its biomechanical qualities have not been completely evaluated. Flexibility tests were performed on 10 calf lumbar spines (L2-L5). Specimens were subjected to pure moments up to 8.2 Nm in flexion, extension, lateral bending, and extension while the resulting movements of L3 and L4 were measured by a three-dimensional motion analysis system. The tested cases included (1) intact, (2) pedicle screw fixation without transfixation after total removal of the L3-L4 disc, (3) pedicle screw fixation with diagonal transfixation, and (4) pedicle screw fixation with horizontal transfixation. Three-dimensional finite element models of the tested surgical constructs were also developed by use of three-dimensional beam elements to investigate the effect of diagonal transfixation and horizontal transfixation on the construct stability and the corresponding stress changes in the screws. When compared with no transfixation, horizontal transfixation significantly improved the lateral bending and axial rotation stability by 15.7% and 13.9%, respectively, but there was no improvement of stability in flexion and extension. By contrast, diagonal transfixation significantly improved the flexion and extension stability by 12% and 10.7%, respectively, but not the lateral bending and axial rotation stability in comparison with no transfixation. Comparison between horizontal transfixation and diagonal transfixation showed that the stabilizing effect of diagonal transfixation was greater in flexion and extension (13% and 11%, P < 0.01) than that of horizontal transfixation but smaller in lateral bending (11%, P < 0.05) and axial rotation (6.6%, P > 0.1). Finite element model predictions of the motion changes were similar to the changes observed in flexibility tests. In horizontal transfixation, the load changes, compared with no transfixion, were a 0.02% increase in flexion-extension, a 27.5% increase in lateral bending, and a 58% decrease in axial rotation, and the magnitudes of the moments applied on both the right and left pedicle screws were identical. However, when diagonal transfixation was achieved by connecting the left superior screw and the right inferior screw, the loads in the left screw were increased by 11.5% in flexion-extension, 43.6% in lateral bending, and 7.9% in axial rotation, whereas the loads in the right screw were decreased by 10.9% in flexion-extension, increased by 0.06% in lateral bending, and decreased by 18.1% in axial rotation. The results of this study showed that diagonal transfixation provides more rigid fixation in flexion and extension but less in lateral bending and axial rotation in comparison with horizontal transfixation. Furthermore, greater stresses in the pedicle screws were predicted in the diagonal transfixation model. These limitations of diagonal transfixation should be considered carefully for clinical application.