Blinded comparison of 5 methods to diagnose atlanto-occipital dislocation (AOD) on plain radiographs and computerized tomography (CT) of the cervical spine. To determine the best method to diagnose AOD. Several methods are proposed for the diagnosis of AOD, including the Power's ratio, X-line method, basion-dens interval, condylar gap, and Harris method. No blinded comparison of the results of these methods has been compared to patient outcome, and there is no information available regarding the accuracy of these methods applied to CT scans. Plain lateral radiographs and CTs of the cervical spine were reviewed in 104 patients, including 6 with AOD. Images underwent a blinded review by a board certified neurosurgeon (D.K.R.), orthopedist (P.A.A.), radiologist (J.C.), and emergency physician (D.B.B.). Each diagnostic method for AOD was applied for determination of sensitivity, specificity, and positive and negative predictive values. The ability to identify relevant anatomic landmarks was also tabulated. Average values for sensitivities, specificities, positive and negative predictive values for each method applied to plain radiographs are: 0.4625-1.0, 0.8933-0.9725, 0.2775-0.45, and 0.975-1.0, respectively. These values for each method applied to CT scans are: 0.7075-1.0, 0.8725-0.9775, 0.3175-1, and 0.98-1.0, respectively. Identification of relevant anatomic landmarks occurred 99.75% of the time when these methods were applied to CT scans compared to 39% to 84% of the time on plain radiographs. Sensitivity, specificity, positive and negative predictive values of these methods improve when applied to CT scans because of better visualization of anatomic landmarks. This result suggests CT scans of the cervical spine may be warranted in all trauma patients suspected of having cervical spine injury.