The authors measured and compared the pre-, intra-, and postoperative three-dimensional shape of the spine during corrective surgery to quantify the specific contribution of positioning, anesthesia, surgical exposure, surgical instrumentation, and postural adaptation of the thoracic and lumbar spine. In 58 adolescent girls with idiopathic scoliosis undergoing corrective surgery by a posterior approach, the three-dimensional geometry of the thoracic and lumbar spine was documented in the standing position before and after surgery using a three-dimensional reconstruction technique based on multiplanar radiography, and the intraoperative three-dimensional geometry was measured using a three-dimensional magnetic digitizer before and after installation of the first rod. Prone positioning, anesthesia, and surgical exposure are responsible for a considerable decrease in all curves in the frontal and sagittal plane. Instrumentation with the first rod produces additional substantial and favorable three-dimensional changes with partial restoration of the normal sagittal curves and sagittal shift of the plane of maximum deformity. Although no loss of correction was observed in the frontal plane when patients resumed their standing position, a "spring-back" effect on the spine was noted in the sagittal plane and a loss of three-dimensional correction was seen in the orientation of the plane of maximum deformity. Surgeons can use the knowledge of these various changes to achieve better results by more careful attention to the preoperative positioning of patients and to curve correction in the sagittal plane when instrumentation is applied to the spine.