There are many kinds of instrumentation systems for posterior operation in the treatment of scoliosis. Cotrel-Dubousset (C-D) system is most widly used for its excellent correction potential and stability. However there were some problems in C-D hook system such as hook dislodgement and correction loss. So, in order to reduce these problems we use transpedicular screw system and compare the results between two systems. We studied 44 cases of scoliosis ( hook 19 cases, screw 25 cases) who were operated with C-D instrumentation from February 1988 to August 1995. The average follow-up period was 54 months in hook group and 23 months in screw group. 1. Operation time was 241 minutes in hook group and 223 minutes in screw group. Average amount of transfusion was 5.0 pints in hook group and 4.6 pints in screw group. 2. Involved segments of main curvature were 7.0 in hook group and 6.6 in screw group. 3. Scoliotic curve was changed from 49degrees to 13degrees (73%) in hook group and from 47degrees to 12degrees (74%) in screw group. Loss of correction during follow up period was 7degrees in hook group and 3 in screw group. 4. Thoracic kyphosis was changed from 24degrees to 26degrees in hook group and from 27degrees to 30degrees in screw group. Lumbar lordosis was changed from 26degrees to 29degrees in hook group and from 26degrees to 31degrees in screw group. 5. Correction rate of rotation of apex vertebrae by Pedriolle method was 43% (from 20degrees to 12degrees) in hook group and 50% (from 22degrees to 11degrees) in screw group. 6. Complications were two cases of hook dislodgement, one delayed deep infection and four cases of progression of curvature in hook group and one case of malinsertion of screw and two cases of progression of curvature in screw group. In conclusion, these results suggested that screw system is more effective than hook system on rotational correction of apex vertebra and prevention of loss of correction.