OBJECTIVE The objective of this study was to compare postoperative sagittal alignment among 3 rod constructs with different rigidity profiles. METHODS This was a dual-center retrospective cohort study involving 2 consecutive cohorts in which patients were surgically treated for adolescent idiopathic scoliosis. Lenke Type 5 curves were excluded. Patients were operated on with all-pedicle screw instrumentation using 3 different rod constructs. The first group was operated on using a hybrid construct (HC) consisting of a normal circular rod on the convex side and a beam-like rod (BR) on the concave side. The second group was operated on with a standard construct (SC) using bilateral BRs in the full length of the fusion. The third group was operated on with a modified construct (MC). The modified rods have a beam-like shape in the caudal portion, but in the cranial 2 or 3 fusion levels the rod transitions to a circular shape with a smaller anteroposterior diameter. Radiographs were analyzed preoperatively and at the first postoperative follow-up (range 1-8 weeks). The primary outcome was pre- to postoperative change in thoracic kyphosis (TK), and the secondary outcome was the ability to achieve postoperative TK within the normal range. RESULTS The HC, SC, and MC groups included 23, 70, and 46 patients, respectively. The 3 groups did not differ significantly in preoperative demographic or radiographic parameters. The mean ± standard deviation of the preoperative main curve was 60.7° ± 12.6°, and the mean of curve correction was 62.9% ± 10.4% with no significant difference among groups (p ≥ 0.680). The groups did not differ significantly in coronal balance or proximal or thoracolumbar curve correction (p ≥ 0.189). Mean postoperative TK was 23.1° ± 6.3°, 19.6° ± 7.6°, and 23.4° ± 6.9° in the HC, SC and MC groups, respectively (p = 0.013), and the mean change in TK was -3.5° ± 11.3°, -7.1° ± 11.6°, and 0.1° ± 10.9°, respectively (p = 0.005). The MC group had significantly higher postoperative TK and less loss of TK compared with the SC group (p ≤ 0.018). Postoperative TK ≤ 10° was found in 12 patients (17%) in the SC group, 1 patient (5%) in the HC group, and 1 patient (2%) in the MC group (p = 0.021). There were no differences in proximal alignment, thoracolumbar alignment, or sagittal vertical axis (p ≥ 0.249). Lumbar lordosis was 58.9° ± 11.2° in the HC group versus 52.0° ± 1.3° and 55.0° ± 11.0° and the SC and MC groups, respectively (p = 0.040). CONCLUSIONS In the 3 rod constructs with different rigidity profiles, significantly better restoration of kyphosis was achieved with the use of bilateral modified rods compared with bilateral standard rods. In the MC and HC groups, the rate of severe postoperative hypokyphosis was significantly lower than in the SC group. This is the first study to describe the clinical use of a rod with a reduced proximal diameter and show marked radiographic improvement in sagittal alignment.