Fifty-six mature beagles underwent lumbar spine destabilization, followed by fusion using four techniques. Spinal cord neuropathologic analysis was carried out to determine the number of abnormalities within each group. Group I (n = 14) had posterolateral bone grafting without instrumentation. Group IIa (n = 14) had Cotrel-Dubousset (CD) pedicle screws and rods. Group IIb (n = 14) had Steffee pedicle screws and plates. Group III (n = 14) had sublaminar wires and rods. All of the animals remained clinically neurologically normal throughout the 6 months of the study. The incidence of moderate to severe neuropathologic changes was 21% in Group I, 18% in Group II, and 64% in Group III. Thus, a significantly higher percentage of neuropathologic abnormalities occurred with sublaminar instrumentation than with no instrumentation (p = 0.027), or with transpedicular instrumentation (p = 0.027). In this controlled animal study, the theoretical advantage of pedicle screws, which should not violate the spinal canal, over sublaminar devices, which must enter the canal, was confirmed.