The Rib Vertebra Angle Difference (RVAD) as defined by Mehta (1972) is used to predict the progression of early onset scoliosis. No clear physical significance has been established for this measurement. The purpose of this study was to evaluate the RVAD along the thoracic spine and the equivalent measurement on 3D reconstructions of the spine and rib cage of early onset scoliosis patients in order to determine their relationship with the geometry of the chest wall and evolution along the spine. The RVAD was measured on PA radiographs of 42 infantile scoliotic patients (Cobb >20°) from T4 to T10 according to the method described by Mehta. The RVAD 3D was computed using the same landmarks from the 3D reconstruction generated from the calibrated biplanar radiographs. Cases were divided into Phase I and Phase II using Mehta's classification based on the rib head overlap with the apical vertebral body on coronal plane radiographs. A linear relationship exists between the Metha (2D) and 3D RVAD for both Phase I (r = 0.87) and Phase II (r = 0.78) patients. For more severe deformities (RVAD 3D ≥ 35°), a relationship was found between RVAD 3D and the axial rotation of the thoracic vertebrae (r = 0.51) in Phase II patients. However, no significant relationship exists between axial rotation and RVAD 3D for Phase I patients as well as Mehta's RVAD. Maximal RVAD measurements were located 2 1/2 levels above the apical vertebra. Results indicated that RVAD 3D provides additional information to Mehta's RVAD on the torsional nature of the deformity. Considering the importance of clinical indices to assess the progression of early onset scoliosis, this study raises some questions on looking solely at the RVAD measured on radiographs at the apical vertebra of Phase I patients and suggests considering also levels above the apex of the scoliotic curve and 3D measurements. Further investigation is required to fully understand the 3D nature of the spine and rib cage deformities.