Four different techniques for posterior atlantoaxial fusion were tested in vitro: 1) wire fixation with one median graft (Gallie type); 2) wire fixation with two bilateral grafts (Brooks type); 3) transarticular screw fixation (Magerl); and, 4) two bilateral posterior clamps (Halifax). The experiment was designed to determine the immediate three-dimensional stability of the spinal construct. Ten fresh human cadaveric specimens were tested intact, injured, and instrumented with each of the fixation techniques. The injury consisted of a severe soft tissue injury model, in which the alar, transverse, and capsular ligaments were transected. The three-dimensional motions of C1 relative to C2 were measured as the specimens were subjected to loads of pure moments in flexion-extension, axial rotation, and lateral bending. Each fixation technique significantly decreased motion in all directions, as compared to the intact and injured spines. We found that the Gallie system generally allowed significantly more rotation in flexion, extension, axial rotation, and lateral bending than the other three fixation techniques. There was generally no significant difference between the amount of rotation with the other three fixation techniques. However, the Magerl technique tended to allow the least rotation. The anteroposterior translation of two points on C1 were about equal for all fixation techniques.