Preoperative curve assessment is important in adolescent idiopathic scoliosis (AIS). Our objective is to clarify the role of side-bending radiographs (SBR) and fulcrum-bending radiographs (FBR) in predicting postoperative Cobb angle in nonstructural and structural curves. Twenty-five consecutive patients with AIS who underwent correction surgery were included. The Cobb angles of structural and nonstructural curves were determined. Cobb angles were measured based on pre- and postoperative standing anteroposterior radiographs of the whole spine. The Cobb angles of SBR and FBR were measured preoperatively. The difference between the Cobb angle at each bending and the preoperative Cobb angle was defined as the predicted correction angle, whereas the difference between the preoperative Cobb angle and postoperative Cobb angle was defined as the surgical correction angle. The correction index was calculated by dividing the surgical correction angle by the predicted correction angle. The difference between the predicted correction angle and surgical correction angle was defined as the prediction error. We compared SBR and FBR for both structural and nonstructural curves in these terms. For both curves, the predicted correction angle of FBR was significantly higher than that of SBR, and the correction index of FBR was significantly lower than that of SBR. Patients with a correction index close to 1 and small prediction error had undergone FBR in the structural curve and SBR in the nonstructural curve. FBR is predictive of postoperative correction angle of the structural curve, whereas SBR is predictive of postoperative correction angle of the nonstructural curve.