Although cervical orthoses are frequently used in prehospital stabilization and in the definitive treatment for lesions of the cervical spine, there is little information about the control of extension-flexion, lateral bending, and rotation given to individual segments by different designs. In an experimental in vitro study with four fresh frozen cadavers, the halo vest was compared with the soft collar, prefabricated Minerva brace, and Miami J collar. The controlling effects for the segments C1-2 and C2-3 were tested for all four devices in the intact and the unstable spine with an Anderson type II fracture of the odontoid. All four orthoses reduced the range of motion at both C1-2 and C2-3 of the intact spine significantly, although none of the three semirigid devices provided a halo-like immobilization in the intact spine. The osteotomy of the odontoid increased the range of motion in the segment C1-2. The soft collar did not give any clinically relevant stability to the unstable spine. Miami J and Minerva brace provided a similar moderate control in the sagittal plane but a much better control of "torque" in the upper cervical spine. The halo vest did not allow any measurable motion in any plane with our experimental external loading. The halo vest seems to be the first choice for conservative treatment of unstable injuries of the upper cervical spine, although pin track problems, accurate fitting of the vest, and a lack of patient compliance lead to clinical failures.