The purpose of this study was to search for constant and reliable reference points on the computed tomographic scans of scoliotic spines to measure vertebral rotation. Several methods using different sets of reference points on computed tomographic scans were tested for reliability. Two methods that gave minimum variation between the readings are presented here with interobserver and intraobserver reliability on 17 girls with adolescent idiopathic scoliosis. The selection of reference points for the two methods were as follows: in method 1, the vertebral rotation was an angle formed by two lines; one, a line joining the junction of the inner surfaces of two laminae and the midpoint of the posterior surface of the vertebral body and second, the vertical plane of the computed tomographic machine. In method 2, three diatum points were marked; one at the junction of the inner surfaces of the lamina, and the other, one each at the junction of the inner surgaces of the lamina and the pedicle. A line bisecting this angle is drawn by the computer and the vertebral rotation was an angle between this line and the vertical plane. The 95% confidence intervals of these two methods for intraobserver variation were between 1–2 to 4.4 degrees. There was no significant difference between the readings obtained by each observer except on three ocasions for observer 1. When these methods were tested for interobserver reliability, method 1 showed significant statistical differences between the readings obtained by the two observers. However, the readings obtained using method 2 were not signifiantly different between the two observers (95% confidence intervals = 3.2 to 5.8 degrees]. The authors recommend the use of Method 2.