In a 5-year prospective study, computed tomographic (CT) morphometry of the lumbosacral vertebral canal was performed on 42 large-breed dogs (21 controls and 21 dogs with lumbosacral stenosis). Dogs were allotted to 4 groups. Group 1 (n = 13) consisted of cadaver specimens obtained from dogs that died or were euthanatized for reasons unrelated to the spine; group 2 (n = 8) consisted of live dogs with no history of clinical signs related to the spine and with normal neurologic examination findings; group 3 (n = 10) consisted of dogs with surgically confirmed lumbosacral stenosis; and group 4 (n = 11) consisted of dogs with suspected lumbosacral stenosis that were managed conservatively. The CT scans were performed, using 5-mm contiguous slices obtained perpendicular to the vertebral canal, from the midbody of the 5th lumbar vertebra through the caudal endplate of the sacrum (L5-S3). Lumbosacral lordosis was minimized in all dogs by positioning them in dorsal recumbency with the hind limbs flexed. A tuberculin syringe calibration phantom was placed within the scanning field of view, parallel to the axis of the spine. In each dog, 11 CT slice locations within the lumbosacral spine were evaluated. At each slice location, sagittal plane diameter, dorsal plane diameter, and transverse area of the vertebral canal, vertebral body, and calibration phantom were measured, using the CT computer's software programs for distance and area calculation. Window/level settings were constant, and all measurements were made by the same operator (JCJ). Accuracy of calibration phantom CT measurements was 100% for sagittal and dorsal plane diameter and was 85% for transverse area.(ABSTRACT TRUNCATED AT 250 WORDS)