Basic reflections on biomechanical tests of spinal implants are reviewed. Four different pedicle screw systems are compared during loading in extension, flexion, side-bending, and rotation. The thoracolumbar and the lumbar spine were tested. Stability against rotation forces was lowest in all systems. Because of this, a cross-connector is recommended in highly unstable situations. All tests were done in cadaver spines with normal bone mineral density (measured before the tests). The first series of tests was done without destroying the cadavers. A second series was done with maximal load until the implants or the bony bed of the screws failed. The results show that rigid implants more often led to pedicle fractures. An unstable screw rod interface led to implant failure. Cadaver tests can provide answers about the primary stability characteristic only. Fatigue tests are not possible in cadaver spines because they become autolytic before enough cycles can be applied.