Retrospective comparative study. This study compares supine vs bending flexibility radiographs and evaluates their ability to predict residual postoperative lumbar curvature after selective thoracic fusion for Lenke 1 and 2 curves across different lumbar modifiers (A, B, and C) in adolescent idiopathic scoliosis (AIS). This was a retrospective review of AIS Lenke 1 and 2 patients who underwent posterior fusion. All patients had preoperative flexibility radiographs including side-bending and supine posteroanterior (PA) films, in addition to pre- and post-operative standing PA and lateral radiographs. We used SurgiMap 2.0 software for all radiographic measurements. Pearson correlations and linear regression models were developed in SAS. A total of 86 patients were included mean age 14.9years and follow-up 72.3 months. Preoperative supine lumbar Cobb angle and preoperative side-bending Cobb angles had similar, positive correlations with postoperative lumbar Cobb angle, r = .55 (P < .001) and r = .54 (P < .001), respectively. Three regression models were built to predict postoperative lumbar Cobb angles from preoperative information: Model S (R2 = .39) uses preoperative supine lumbar curve; Model B (R2 = .44) uses preoperative side-bending lumbar curve; Model SB (R2 = .49) uses both preoperative supine and side-bending lumbar curves. Model S and B performed just as well as Model SB. Either supine or side-bending radiographs alone may be used to estimate mean residual postoperative lumbar curvature after selective posterior thoracic fusion, but little is to be gained by taking both supine and side-bending radiographs.