Many salvage options for failed thoracic pedicle screws exist including the use of a different trajectory or the augmentation of the screw with polymethylmethacrylate cement. Although polymethylmethacrylate immediately increases the construct stiffness and the pull-out strength, it may cause bone necrosis, toxin relaxation, and/or neural injury. On the other hand, calcium sulfate bone grafts have a high potential for biologic incorporation and no thermal damage effect. In the current study, polyaxial pedicle screws were first inserted with a straightforward approach on both sides in 17 fresh human cadaveric thoracic vertebrae. The maximal insertion torque for each screw was measured and then the pull-out strengths were recorded. Afterward, these pedicle screws were randomly assigned to be replaced either by graft augmentation or by anatomic trajectory technique for salvage. The graft-augmented screws were placed using the previous holes. The maximum insertional torque for each anatomic trajectory screw was measured. Finally, the pull-out strengths of the revision screws were recorded. The mean maximum insertional torque decreased with the anatomic trajectory salvage technique when compared with the straightforward approach, 0.23 versus 0.38 Nm, respectively (P=0.003). The anatomic trajectory revision resulted in decreased pull-out strength when compared with the pull-out strength of the straightforward technique, 297 versus 469 N, respectively (P=0.003). The calcium sulfate graft augmentation increased the pull-out strength when compared with the pull-out strength of the straightforward technique, 680 versus 477 N, respectively (P=0.017). The mean pull-out strength ratio of revised screw to original was 0.71 for anatomic trajectory and 1.8 for graft-augmented screws, a statistically significant difference (P=0.002).