BACKGROUND: Lumbar pedicle screws are placed for internal fixation and help to enhance bony fusion. Optimal screws are medially directed, should be parallel or pointing to the superior endplate, and penetrate 50%-80% of the vertebral body. "Nonparallel" pedicle screws can be inadvertently placed within the confines of the pedicle and vertebral body but are sometimes replaced to obtain a more acceptable postoperative image. A nonparallel (suboptimal) screw is one that is located within the pedicle and body and does not violate bone; however, it is not parallel to the superior endplate. These "cored-out" grooves left in the bone from the initial tap and screw placement may compromise the integrity of the bone and the construct. METHODS: Dual-energy x-ray absorptiometry scans and L4-5 laminectomies were performed on 6 fresh-frozen cadaveric lumbar spines. We placed 2 optimal pedicle screws in L4, 1 optimal screw in L5, and 1 suboptimal screw in L5 (construct A). Axial rotation, flexion/extension, and lateral bending were tested. The suboptimal screw was repositioned in an optimal trajectory and retested (construct B). Pullout strength was performed on optimal and revised L5 pedicle screws. RESULTS: The mean axial rotation stiffness was 1.31 N-m/degrees ± 0.22 in construct A and 1.19 N-m/degrees ± 0.17 in construct B (P = 0.023; 95% CI [CI], 0.20-0.02). The mean lateral bending stiffness was 0.015 N/mm ± 0.002 in construct A and 0.016 N/mm ± 0.002 in construct B (P = 0.3; 95% CI, 0.0008-0.001). The mean flexion/extension stiffness was 0.0139 N/mm ± 0.002 in construct A and 0.0126 N/mm ± 0.002 in construct B (P = 0.01; 95% CI, 0.002-0.0004). Axial rotation and flexion/extension stiffness were significantly different between the 2 groups. The mean pullout strength was significantly higher in the nonrevised parallel screw group compared with the reimplanted parallel screw group (906.93 N ± 271.17 vs. 608.32 N ± 207.23, P = 0.031). Dual-energy x-ray absorptiometry imaging demonstrated 4 osteopenic and 2 osteoporotic specimens, although differences in bone mineral density did not play a significant role in assessing either the biomechanical parameters or the pullout strength. CONCLUSIONS: Great care is warranted in the initial placement of lumbar pedicle screws. Revising a nonparallel screw placement decreases pullout strength and alters biomechanical movements (axial rotation and flexion/extension) in patients with decreased bone mineral density. If a screw is inadvertently placed nonparallel to the endplate but is within the confines of the pedicle and vertebral body with adequate bone purchase, it should not be revised and rather be left in its place.