Study Design. The aim of 1 his study Is to quantity cliffs run ces between three dimensional rotations in space and their calculated values, either on two-dimensional projections (radiographs or computed tomographic scans) or three-ctimensicnal calculations using various mathematical procedures. Objective. To use a vertebral model to quantify differences between three-dimensional rotations and their calculated values, using two-dimensional projections or various three-dimensional mathematical procedures. Methods. A specific program allowed us to move a geometric vertebral model in space using given values and sequences of lateral, sagittal, and axial rotations. Differences in positions due to differen t sequences were visualized and quantified. Differences due to rotation around global or vertebral axes were considered. Results. For rotations of about 10º, differences are about 2º between three-dimensional and projected angles. Differences increase when combined rotations are large, as generally occurs in a scoliotic spine. They reach 16º for lateral and sagittal rotations of 30º. Conclusion. Axial rotation measured on transverse projection is misleading for vertebrae rotated in space. Moreover, dealing with large three-dimensional rotations is meaningful only if the used mathematical convention is given.