Outcome from surgery for cervical spondylosis is often disappointing. To identify possible causes of poor outcome 56 such patients referred for post-operative computed myelography or MRI were evaluated, 22 of which eventually had further surgery. Alternative diagnoses to cervical spondylosis were eventually established in 14.3 %; 26.8% had spinal cord atrophy 15.6% of which also had myelomalacia; 28.6% had diffuse spinal canal stenosis; and in 57.1% surgery had failed to decompress the spinal canal. These findings can be partly explained by patient selection criteria; nevertheless they do serve to emphasises the point often ignored in discussions of the efficacy of surgery in cervical spondylosis, that operations significantly often fail to achieve adequate decompression. Furthermore there was no evidence in this material that osteophytes regress after spinal fusion.