There have been many studies on adolescent idiopathic scoliosis related abnormal muscle contractions of the spine. However, previous studies using surface electromyography to investigate paraspinal muscle asymmetry are controversial, lacking in clarity of results, and hindered by methodological limitations. The purpose of this study was to investigate the relationship between imbalance factors including surface electromyography activity according to the scoliosis curve type and leg length discrepancy and adolescent idiopathic scoliosis curve types. Seventy-nine patients with scoliosis were prospectively enrolled and were divided into five types: single thoracic, thoracolumbar, lumbar, double thoracic, and double major. Cobb angle and structural variables were measured. Surface electromyography examinations were conducted at the 7th, 12th thoracic erector spinae, 3rd lumbar erector spinae, and multifidus muscles during the superman position keeping prone spinal extension to lift the arms and legs off the floor. Whole spine radiographs were obtained to measure the Cobb angle, coronal imbalance, pelvic height and angle, and femoral head height. In the double major, thoracolumbar, and lumbar types, the mean root mean squared (RMS) EMG amplitudes were significantly higher on the convex side than the concave side (P < 0.005). In the DM type, the mean RMS EMG amplitudes of EST7 and ESL3 where the apex was located were significantly higher at the convex side than those of the concave side (P < 0.005, effect size (Cohen’s d) for EST7/ESL3: 0.517/0.573). The TL and L types showed a similar pattern. The mean RMS EMG amplitudes of the EST12 concave side and MuL3 and ESL3 concave sides were significantly lower than those of the convex side in the TL and L types, respectively (P < 0.008, effect size (Cohen’s d) for EST12/MuL3/ESL3: 0.960/0.264/0.448). Conversely, there were no significant differences in the single thoracic and double thoracic types. All structural variables (coronal imbalance, pelvic height and angle, and femur head height) were higher in the lumbar type, but only coronal imbalance was significantly different (P < 0.05). Different patterns of asymmetry of paraspinal muscles and structural variables were described based on the curvature of the spine. L type showed that EMG activity was asymmetric in the paraspinalis muscles where the apex was located and that structural asymmetry, such as coronal imbalance was significantly greater than other types. DM type showed similar paraspinalis asymmetry pattern to the ST type but there was no structural asymmetry in DM and ST types. TL type has the features of both thoracic and lumbar origins. Understanding these could contribute to the management in correcting scoliosis.