A technique for simultaneous roentgen stereophotogrammetry (RS) was developed, and its accuracy was assessed. In vitro models fabricated from dried cadaveric C4 and C5 vertebrae were used to simulate the motion behavior of the cervical spine. Metallic markers made of Vitallium beads (diameter < 0.3 mm) were implanted into the posterior and anterior surfaces of each vertebra at surgically accessible locations to simulate the bead placement for both posterior and anterior surgical approaches to the cervical spine. A series of roentgen stereo pairs were obtained to systematically assess the accuracy (validity) of displacement measurements in anteroposterior (AP) translation, axial rotation, and flexion/extension. In addition, the effects of soft tissue density on the accuracy of the system were investigated by obtaining a series of roentgen stereo pairs with the experimental model immersed in a water bath. The coordinates of the metallic markers on the radiographs were then digitized by two raters who were not informed of the actual motion (i.e., blind study). The results indicated a high accuracy throughout the study. Overall root mean square errors were 0.07 mm for AP translation, 0.08 degrees for axial rotation, and 0.14 degrees for flexion/extension. The corresponding accuracy estimates (R2 values by linear regression analysis) were very high (0.992, 0.998, and 0.995) when the measurement results were compared with the actual displacements. The water bath did not affect measurement accuracy, indicating that soft tissue density should have little effect on the accuracy of the technique for in vivo applications. This system appears to be an accurate and reliable method for assessment of simulated in vivo cervical spine motion, regardless of the rater. The technique has been further used in in vivo assessment of cervical spine kinematics in one patient to confirm the efficacy of the developed technique.