Obtaining accurate measurements of scoliosis from two-dimensional (2-D) radiographs can be challenging because of the three-dimensional (3-D) nature of the deformity. Previous studies have shown that the sagittal plane, in particular, is misrepresented on 2-D radiographs because of the influence of axial plane rotation. The purpose of the current study was to define a methodology for measuring the 3-D segmental sagittal alignment of the spine in patients with adolescent idiopathic scoliosis (AIS) and to assess the effect of axial plane rotation on differences between 3-D and 2-D measures of deformity. Preoperative and postoperative EOS images of 120 consecutive patients with AIS (primary thoracic curves) treated with segmental thoracic pedicle-screw instrumentation were analyzed in the "3-D sagittal plane." The technique measured 3-D kyphosis or lordosis in the specific plane of sagittal motion for each spinal motion segment. The kyphosis (+) and lordosis (-) values of the segments from T5 to T12 were summed to give the 3-D measurement of T5-T12 kyphosis. These values were compared with the standard 2-D measurements of T5-T12 kyphosis on lateral radiographs, and a correlation analysis with regard to axial plane rotation of the apex was performed. The average age (and standard deviation) of the patients was 14 ± 2 years. The mean preoperative Cobb angle on the standard 2-D view was 55° ± 10° and on the 3-D view was 52° ± 9° (p ≤ 0.001). On the 3-D view, the mean preoperative T5-T12 kyphosis was 6° ± 14°, and the kyphosis significantly increased to 26° ± 6° postoperatively (p < 0.001). The T5-T12 kyphosis on the standard 2-D view measured 18° ± 13° preoperatively and 27° ± 6° postoperatively (p < 0.001). The difference between the 2-D and 3-D measurements of T5-T12 kyphosis strongly correlated with apical vertebral rotation (r = 0.85; p < 0.01). Routine 2-D measurements of thoracic kyphosis erroneously underestimate the preoperative loss of kyphosis in AIS because of errors associated with axial plane rotation, an inherent component of thoracic scoliosis.