In a random, controlled laboratory study, pedicle screws placed in human cadaveric vertebrae were axially loaded to failure. Three repair methods were tested. Use of low pressure polymethylmethacrylate yielded 149% of the original pullout strength, milled bone yielded 70% of the initial pullout strength, and matchstick bone yielded 56% of the initial pullout strength. Two incidents of cortical penetration during reinsertion in the matchstick group occurred. The results were statistically analyzed.