Three hundred and fifty-two patients had a one-stage posterior spinal arthrodesis between 1960 and 1984 using one of four types of instrumentation: a Harrington distraction rod, Harrington distraction and compression rods, Harrington distraction and compression rods with a device for transverse traction, and a Harrington distraction rod with sublaminar wires. All of the patients were female (age-range, eleven to nineteen years), and all had idiopathic scoliosis with a single right or double thoracic curve. The minimum length of follow-up was two years. No significant difference was found among the four groups relative to the amount of correction that was obtained at operation or maintained two years after operation. An average of 13.5 per cent of correction was lost during follow-up in the patients who were treated with postoperative immobilization, and an average of 27 per cent was lost in the patients who were treated with sublaminar wires without immobilization. The use of a straight Harrington rod reduced normal thoracic kyphosis, the addition of a compression rod corrected hyperkyphosis, and the use of a rod with sublaminar wires corrected thoracic hypokyphosis or thoracic lordosis.