Of thirty-one patients who underwent posterior fusion for congenital scoliosis from 1972 through 1977 at the Children's Hospital Medical Center, Boston, Massachusetts, and were followed for two years or more, eighteen (average age, fourteen and one-half years and average curve, 62 degrees) were treated by spine fusion using Harrington instrumentation, and thirteen (average age, and one-half years and average curve, 43 degrees) were treated by fusion without instrumentation. Correction of the curve in the instrumented group was obtained at operation, while in the group without instrumentation correction was attempted using a plaster jacket applied during the postoperative period. After an average follow-up of thirty-four months in the group with Harrington instrumentation, the average curve was reduced from 62 to 40 degrees, for an average correction of 22 degrees, while in the non-instrumented group, after an average follow-up of fifty-three months the average curve was reduced from 43 to 38 degrees, for an average correction 5 degrees. A myelogram using water-soluble contrast medium should be performed in all patients who are to have instrumentation and in all patients who have any neural abnormality or are suspected of having diastematomyelia. An intraoperative wake-up test was used in all patients who had instrumentation after 1973 and averted neural complications in one. Proper treatment of congenital scoliosis requires early recognition of curves that have already progressed or will certainly do so. Fusion without instrumentation then is sufficient. If correction is necessary, staged procedures (halo-femoral traction, anterior release, and posterior fusion) may be required for severe curves. For the less severe curves, instrumentation as the primary means of obtaining correction proved to be safe and effective in this small series, but should only be attempted by experienced surgeons in institutions with all of the necessary facilities.