Sixteen fresh-frozen spines from cadavers (C4-T1) were randomized on the basis of dual energy xray absorptiometry analysis of bone mineral density. The specimens were subjected to physiologic loads (<or= 1.5 N-m) in a device that applied pure unconstrained, flexion and extension, lateral bending, and torsional moments. After a major injury, including a wide C6 laminectomy, bilateral capsulectomies, and 65% facetectomy at C6-C7, four constructs were applied to each specimen in a balanced sequence, and the stability tests were repeated. All the constructs were identical posterior cervical rods secured to the spine with lateral mass screws in C5 and pedicle screws in C7. The constructs differed only in the presence or absence of a transverse connector between the rods, presence or absence of lateral mass screws in C6, and unicortical or bicortical lateral mass screws. Insertional screw torque was higher in specimens with greater bone mineral density. Pedicle screws had greater torque than lateral mass screws. Posterior cervical spine rod fixation provided an equivalent stability by use of either unicortical or bicortical lateral mass screws. A transverse stabilizer between the rods reduces the number of lateral mass screws needed in posterior cervical spinal rod systems. Repeated stability tests, even when done with physiologic loads, may compromise the screw-bone interface. This effect is most pronounced in specimens that have low bone mineral density.