METHODS In a retrospective study, the incidence of false positive and false negative interpretation of x-rays for solid spinal arthrodesis with spinal instrumentation was evaluated in 75 patients. OBJECTIVE To evaluate the accuracy of the interpretation of x-rays for diagnosing solid spinal arthrodesis in patients with spinal instrumentation. BACKGROUND This retrospective study compared spinal fusion, as determined by direct observation and radiographic evaluation, in 75 patients with instrumented lumbar fusions using multiple devices. The fusions included posterolateral fusions or posterolateral with interbody fusions. Autograft, allograft, and a combination of these also were used. METHODS A single blinded examiner reviewed all x-rays immediately before the spinal hardware was removed and the fusion mass was explored by the surgeon. RESULTS There was a positive correlation between x-rays and the observations at the time of surgery in only 68% of the patients. CONCLUSIONS This study indicates that the accuracy of x-ray interpretation for spinal arthrodesis is only 68%. The L4-L5 level was the most difficult level to fuse and the most difficult to interpret using x-rays. Patients with persistent back pain, when nonmechanical causes have been ruled out, should be considered for surgical exploration of the fusion mass even if x-rays appear to indicate a solid fusion.