The purpose of this study is to determine how accuratively and reliably the Perdriollo s method can measure vertebral rotaton according to the level of vertebra and true axial rotation. Standard AP radiographs were taken using 10 dry human vertebra (two set of T3, T6, T9, T12, L3) with 5degrees increments in axial rotation, ranging from 0degrees to 50degrees. In order to evaluate the reliability of Perdriolle s method, three observers measured each radiograph twice, and intet-and intraobserver variance were estimated. The accuracy of Perdriolle's method was analyzed by total error analysis, root mean square error (RMSE) and Deviation of differences. 372 errors were made in the 660 measurement and 88.9% of all measurements were within ± 5degrees of true rotation angle. There were no signigicant difference in the intraobserver measurement at each level whereas interobserver measurements were significantly different only at T3 level. The measurement was also the least accurate at T3 (RMSE=6.2292). We concluded that Perdriolle's method using torsion meter is an accurate and reliable one to measure the vertebral rotation with less accuracy and reliability in high thoracic level.