For young children with scoliosis before growth spurt, suhcutaneous lengthening without fusion was designed by Harrington and modified by Moe and Luque. However, many problems including spontaneous fusion, rod breakage, and hook disloclgement have been ohserved. CotrelDubousset(CD) instrumentation was sometimes used, but it usually resulted in failure due to soft tissue adhesion around the rough surface of ordinary CD rod. We tried to use the smooth CD rod, transvcrse-pedicle clawing on the upper part, and pedicle screw inscrtion on upper and lower part of the curve to reduce the hardware failures. Among 8 patients in whom suhcutaneous lengthening with smooth CD rod was carried out hetween October l992 and Suly 1996. 4 cases perfomed with final spinal fusion were analysed. There were I central core disease, 1 multicore disease and 2 idiopathic scoliosis(infantile and juvenile type). Mean age at the first operation was l0.0(8.8-11.8) years, and the Risser sign was all grade 0 except one with grade 1. Suhcutaneous lengthening was performed every 5 or 6 months Mean lengthening duration was 22(9-39) months and mean age at spinal fusion was 11.7(9.6-13.8) years. Mean Cobb angle decreased from 7ldegrees (55degrees-88degrees) at preoperative stage to 32 (10degrees-59degrees) at the last follow-up. There were 5 complications during 21 operations, and three hardware failures comprised 2 hook dislodgcment and 1 screw pull-out. Crankshaft phenomenon happened in I case who had had a posterior fusion in young age(9.6 years) due to laminar fracture. The suhcutaneous lengthening with smooth CD rod can he another option of treatment for young children with severe scoliosis. prescrving the powth potential of involved vertebrae with few complications.