Thoracic hypokyphosis is common in adolescent idiopathic scoliosis (AIS), making sagittal restoration a primary goal. "Rod bend-back," the loss of rod curvature during correction, often leads to insufficient thoracic kyphosis (TK) restoration. This study aimed to investigate the impact of rod characteristics (manual-bent vs. pre-bent) and correction techniques (rod rotation [RR] vs. vertebral coplanar alignment [VCA]) on the rod bend-back rate and final sagittal outcomes. We retrospectively reviewed 58 patients with AIS (57 females and 1 male; mean age 15.3 years) with Lenke type 1 or 2 curves who underwent posterior fusion between 2011 and 2025. Patients were categorized into three groups: Group M (manual-bent rods and RR, n=36), Group P (pre-bent rods and RR, n=10), and Group C (pre-bent rods and VCA, n=12). Propensity score matching (PSM) was used to compare Group M vs. P and Group M vs. C. The primary outcome was the concave rod bend-back rate, with post-operative TK assessed as a secondary outcome. Multiple regression analysis was performed to identify predictors of rod bend-back. PSM revealed no significant differences in bend-back rates between Group M and P (53.4% vs. 49.3%, p=0.50) or Group M and C (53.0% vs. 41.8%, p=0.16). However, Group C achieved significantly higher postoperative TK compared to Group M (23.5° vs. 17.7°, p<0.01). Multiple regression analysis identified preoperative sagittal spinal rigidity (flexion TK: p=0.02) and the magnitude of the initial rod bending angle (p<0.001) as the dominant predictors of bend-back. These results suggest that rod bend-back is primarily governed by patient-specific sagittal spinal rigidity and the degree of initial bending rather than the rod type or correction technique. While VCA is effective for kyphosis restoration, achieving optimal sagittal alignment in rigid cases requires sufficient spinal release, as over-bending the rod alone may be ineffective.