STUDY DESIGN: A prospective comparative evaluation of the commonly accepted or described radiologic techniques to determine curve flexibility in adolescent idiopathic scoliosis (AIS), comparison of the results to those obtained by supine traction radiographs taken with the patient under general anesthesia (UGA) just before surgery and correlation of all findings to surgical correction. OBJECTIVE: To determine if supine traction radiographs taken with the patient UGA help provide better assessment of curve flexibility and better predicting surgical correction. SUMMARY OF BACKGROUND DATA: Supine lateral bending radiographs are the standard methods of evaluating curve flexibility before surgery in idiopathic scoliosis. Supine traction radiographs have also been used at the authors' institution in addition to the supine lateral bending radiographs before surgery, believing that it is usually more helpful to analyze the response of the main and compensatory curves to corrective forces. METHODS: A total of 34 consecutive patients with AIS who had surgical treatment were studied. Preoperative radiologic evaluation consisted of standing anteroposterior and lateral, supine lateral bending and traction, fulcrum bending radiographs, and also supine traction radiographs taken with the patient UGA just before surgery. All structural curves were measured, and the flexibility ratio was determined on each radiograph. The amount of correction obtained by all radiographic methods was compared with the amount of surgical correction by evaluating the differences from surgery as absolute values. Mean absolute differences from surgery were used to determine the confidence intervals. Statistical differences were calculated with the comparison of the exact 95% confidence intervals for the mean. RESULTS: Curves were accepted to be moderate if between 40 degrees and 65 degrees (29 patients) and severe if >65 degrees (5 patients). In these 29 patients, average frontal Cobb angle of the thoracic and lumbar curves were 49.7 degrees (range 40 degrees-60 degrees) and 39.4 degrees (range 22 degrees-58 degrees), respectively. For the moderate thoracic curves, fulcrum radiographs provided the best amount of flexibility, with no significant difference from traction with the patient UGA but with significant difference from bending radiographs. For the moderate lumbar curves, flexibility obtained by fulcrum and bending radiographs were significantly better than traction radiographs with the patient UGA. For the lumbar and thoracic curves more than 65 degrees, traction radiographs with the patient UGA provided clearly better flexibility compared to bending and fulcrum radiographs, however, the number of patients is not enough to determine whether the differences are statistically significant. Better flexibility in traction radiographs with the patient UGA helped us eliminate the need for anterior release in all 5 patients who had severe and rigid curves more than 65 degrees, which did not bend to less than 40 degrees and were planned to have anterior release. CONCLUSION: Fulcrum higher than bending higher than traction with the patient UGA is the order of radiographs for better predicting flexibility and correction in curves between 40 degrees and 65 degrees. Flexibility obtained at traction radiographs with the patient UGA is clearly better in numerical values, and closer to the amount of surgical correction than the amount of flexibility at fulcrum and side-bending radiographs for curves larger than 65 degrees, although not statistically significant as a result of the small number of patients in this group. However, pedicle screw instrumentation provides even more correction than the traction radiographs with the patient UGA. Thus, traction radiographs with the patient UGA may show much better flexibility, especially in more than 65 degrees and rigid curves.