Cervical spondylosis is common condition rarely associated with radiculomyelopathy which surgical treatment, according to meta-analysis, is not better than nonsurgical. Our hypothesis was that neurodecompression which type is chosen according to spinal alignment should result in better functional improvement comparing with nonsurgical treatment. Between January 1, 1998 and December 31, 2007 a total of 77 patients with spondylogenic myelopathy were selected for the study. The inclusion criteria were symptoms and signs of myelopathy Ranawat grade III. Exclusion criteria were amyotrophic lateral sclerosis (ALS) and multiple sclerosis (MS). The curvature of the cervical spine was determined by Ishihara index. Anterior corpectomies and fusion was performed in the kyphotic spines, laminectomy with fusion in patients with neutral position, and open door laminoplasty in lordotic spines. Clinical improvement was assessed as differences between preoperative and 1-year follow up Nurick, modified Japanese Orthopedic Association (mJOA) myelopathy scales and walking test. Preoperative and postoperative transverse cord area and subarachnoid space were measured. Forty-four male and 31 female patients were surgically treated. Two patients with electrophysiological signs of ALS were excluded. Preoperative and postoperative mean +/- SD mJOA index was 9.15 +/- 1 and 13.01 +/- 1.4 (p < 0.001), Nurick grading scale 3.05 +/- 0.7 and 1.8 +/- 0.6 (p < 0.001), walking time (sec) 64.4 +/- 3.2 and 46.2 +/- 3.3 (p < 0.001), and number of steps 69.7 +/- 4.4 and 57.6 +/- 2.8 (p < 0.001) respectively. Preoperative and postoperative transverse cord area (mean +/- SD, mm2) was 46.7 +/- 5.4 and 60.2 +/- 2.6 (p < 0.001), and subarachnoid space 48.0 +/- 4.9 and 68.8 +/- 8.5 (p < 0.001) respectively. Our results showed that surgical treatment is beneficial for patients with spondylogenic myelopathy.