This study was performed to compare the diagnostic accuracy and sensitivity of computerized tomography (CT), myelo-enhanced computerized tomography (MECT) and magnetic resonace imaging in the diagnosis of spinal stenosis and to determine the prevalence of stenosis in the various regions of the lumbar spine. The materials were 131 levels of 63 patients with spinal stenosis who were treated with posterior decompression, transpedicular screw fixation and posterolateral fusion at Seoul National University Hospital during the period of June 1988 to July 1989. On CT and MECT, central stenosis was defined as anteroposterior dural sac diameter less than 10mm. Lateral recess was considered stenotic when subarticular space at superior articular facet level was less than 3mm. Foraminal stenosis was defined as narrowing less than 3mm at foraminal cut. On MRI, besides the horizontal cuts using the same criteria, stenosis was considered present when there was narrowing of dural sac diameter less than 10mm by reasons other than herniated nucleu pulposus in the central sagittal plane, indentation and encroachment of dural sac or spinal canal at the sagittal plane passing medial to the facet plane showing lateral recess, or distortion of normal tear drop configuration of the intervertebral foramen at the sagittal plane passing lateral to the facet joint. With evaluation by above criteria and comparing them with operative findinga, following results and conclusions were obtained. 1. Combined CT-MRI is the most accurate diagnostic method for spinal stenosis with high overall sensitivity. 2. Combined CT-MRI is most effective in the evalution of the canal, lateral recess and foramen canal. 3. Combined CT-MRI is effective in both bony and ligamentous pathology, but cost-effectiveness should be considered. 4. The most prevalent form of spinal stenosis is the combined central-lateral-foraminal stenosis comprising 64.9%. Stenosis involving only one region was relatively rare (13.7%)..