The results were reviewed for thirty patients in whom a thoracolumbar fracture had been treated with a Harrington rod placed three segments cephalad and two or three segments caudad to the injured vertebra and a short arthrodesis with removal of the Harrington rod after at least twelve months. The patients were evaluated with clinical examinations, lateral radiographs made with the spine in flexion and extension, and questionnaires at a median of thirty-two months after the stabilization. All of the arthrodeses were successful. The twenty patients who had a burst or compression fracture and who had posterior instrumentation lost only an average of 9 degrees of sagittal correction before attaining a solid fusion. For all thirty patients, the average motion in the sagittal plane at the vertebrae that had been spanned by the Harrington rod but not included in the arthrodesis ranged from 9 degrees at the third and fourth lumbar level to 2 degrees at the eighth and ninth thoracic level. The average total motion between five vertebrae that had been spanned by the rod but did not have an arthrodesis was 24 degrees; between four vertebrae, 23 degrees; and between three vertebrae, 19 degrees. In this series of thoracolumbar fractures treated with a longer Harrington rod than needed to span the level of the arthrodesis and a short arthrodesis, there was a documented return of motion to segments that had been previously spanned by the instrumentation but not included in the arthrodesis and good or excellent functional results over-all.