For biomechanical purposes, interbody fusion cages should not dislocate, should provide high stability,and should have a low subsidence risk. Zientek (Marquardt Medzintechnik), Stryker (Stryker Implants),and Ray lumbar interbody fusion cages (Surgical Dynamics) were tested in this study. They were implanted by pairs from a posterior approach without further stabilization. In a first step, each cage design was implanted into four human L3-4 segments and extracted posteriorly under an axial preload of 200 N. In a second step, standard flexibility tests were carried out with 24 human L2-3 and L4-5 specimens in an intact condition, directly after cage implantation, and after cyclic axial compression loading (200-1000 N, 40,000 cycles, 5 Hz). In a third step, a destructive axial compression test was carried out. Maximum pullout force was highest with Ray cages (median 945 N),followed by Zientek (605 N) and Stryker cages (130 N). With all three cage designs, primary stability was higher in lateral bending and flexion than in extension and axial rotation. Implantation of Ray cages caused a decreased range of motion in all three loading directions ranging between 49% and 99%. Zientek cages only stabilized in lateral bending, flexion, and extension (45-78%) and Stryker cages in none of the three loading directions. Cyclic loading caused an increased range of motion in all cases up to 190%. Axial compression force at failure was 8413 N with Ray cages, 8359 N with Stryker cages, and 5486 N with Zientek cages. The cage design seems to influence the dislocation tendency. In this regard,threaded cages or cages with anchorage systems seem to provide more security. The stabilizing effect seems to be mainly influenced by factors such as the degree of distraction or destruction of the facet joints rather than by the cage design.