The Perdriolle torsionmeter assesses vertebral rotation on a spinal radiograph. It is frequently used ot measure improvement in spinal derotation following Cotrel-Dubousset instrumentation for scoliosis. In this study, Intraobserver and intraobserver measurement error was examined during use of the torsionmeter. Intraobserver error was as follows: 53% of the measurements were accurate to within 5°. Error from the actual value averaged 6°. Interobserver error was as follows: Among six observers, only one third of the radiographs had measurements within 5° of each other. Another one third erred bymore than 10°. Because of this significant Intraobserver and Interobserver error, precise measurements of rotation using the torsionmeter cannot be expected. Efforts to quantify spinal derotaton with the torsionmeter after Cotrel-Dubousset instrumentation may not be valid.