surgical outcome and neurologic recovery have improved dramatically (23,29). However, assessment of exact pathologic segment in multi-level CSM is still demanding. Magnetic resonance imaging (MRI) is usually accepted as the primary and best diagnostic tool to evaluate CSM (2,10). It is possible to diagnose neural compression by canal and differentiate an intramedullary lesion simultaneously with a AIM: To compare the accuracy of determining pathologic segment between three-position MRI (3P-MRI) and post-myelographic CT (PMCT) in cervical spondylotic myelopathy (CSM) by assessing the degree of inter-observer and intra-observer agreement. MATERIAL and METHODS: We retrospectively reviewed 3P-MRI and PMCT for the diagnosis of multilevel CSM in 136 patients who underwent surgery. Using an assessment scale, 8 blind observers with various clinical experiences examined 5 parameters: spinal canal narrowing, foraminal stenosis, bony abnormality, intervertebral disc herniation, and nerve root compression. Spinal canal, neural foraminal, spinal cord and disc protrusion diameters were measured. Intra-observer and inter-observer agreement of each image was analyzed. RESULTS: Spinal canal width and foraminal diameter was found to be significantly smaller in 3P-MRI compared to PMCT. No significant differences of cervical cord diameter and the size of disc protrusion measured in 3P-MRI compared to PMCT were observed. Comparing between 3P-MRI and PMCT, disc abnormality and nerve root compression showed better agreement on 3P-MRI, whereas foraminal stenosis and bony lesion showed better agreement on PMCT. CONCLUSION: In the present study, PMCT was still useful in diagnosis of the foraminal stenosis and bony lesion compared to 3P-MRI but showed limitation in disc abnormality and nerve root compression. Even though PMCT may provide valuable additional information in difficult or ambiguous cases, universal standard of 3P-MRI showed higher reliability in detecting pathologic levels in CSM patients.