Structured Abstract Objective (a) To evaluate intra‐ and inter‐observer reliability in landmarks placement along the three planes of space on cone‐beam computed tomography (CBCT) data sets; (b) To evaluate whether the reliability of each landmark differs in CBCT scans characterized by two different voxel dimension and quality. Setting and Sample Population A total of 84 scans were used in this study: 49 scans were taken with the NewTom 3G, and 35 scans were taken with the NewTom 5G. The scans were characterized by an isotropic voxel dimension of 0.36 and 0.30 mm for the NewTom 3G and the NewTom 5G, respectively. Methods A total of 13 landmarks were placed according to the corresponding definitions in 3D, also presented in this study: Foramen Spinosum (R/L), Nasion, Sella, Gonion (R/L), Pogonion, Menton, A point, Anterior nasal spine, Posterior nasal spine, Basion, Cribriform Plate (CR). Intra‐ and inter‐observer reliability and Intra‐class correlation coefficients (ICC) ICC for landmarks identification were assessed. Five reference and registration planes based on the above‐mentioned landmarks were also presented. Results ICC both for the intra‐ and inter‐observer measurements had a score larger than 0.9 in all directions, except in the sagittal direction for CR. Regarding intra‐ and inter‐observer reliability, only N, S, and Ba scored well in all directions. Conclusions Most of the landmarks analysed displayed a high reliability along at least 2 directions. The choice of landmarks to define registration and superimposition planes must be carefully selected, as the reliability of these planes is inherited from the one of the landmarks defining them.