Appropriate levels for instrumentation and fusion in scoliosis have been a matter of debate among surgeons since the introduction of operative management of this deformity. We set out to examine the hypothesis that the amount of correction achieved in all planes during surgical instrumentation of a curve should be less than, or comparable to, the degree of correction attainable at any non-instrumented adjacent curve. An algorithm was designed to facilitate preoperative planning and intraoperative performance of spinal fusion procedures in the management of scoliosis. To test the validity of the hypothesis and the proposed algorithm, measurements were taken from the preoperative radiographs of 200 patients. The dimensions of the curves were obtained from an initial set of four X-ray films: (1) standing anteroposterior film of the whole spine, (2) standing lateral film of the whole spine, (3) two properly performed side-bending films including each curve of the spine. With this data, a plan was designed using the algorithm. The results of this plan were compared with the actual results of the surgery, which were revealed only at this stage. All patients in whom actual instrumentation levels fell within those predicted by the proposed algorithm had no imbalance at follow-up. All patients whose actual instrumentation levels were short of those recommended by the algorithm showed obvious imbalance on final postoperative standing radiograph.