Objective: to compare the accuracy and reliability of orthodontic dental measurements obtained from CBCT models, digital models and conventional plaster models. Material and methods: The sample consisted of plaster dental models of 30 orthodontic subjects. Dental arch measurements, including mesiodistal widths of teeth and arch widths. Were made with the calipers. The patients were also scanned with CBCT, and measurements were made digitally on a 3-dimensional based dental measurements software (On Demand 3D, Cypermed, Seoul, Korea), Plaster models were scanned with a digital scanner (Canon Pixma MX300, Canon, USA) and measurements were made digitally on Onyx Ceph. Software (Ver. 2.7.18, OnyxCeph, Chemnitz, Germany). Results: For the accuracy evaluation, statistically significant differences were found between conventional models and CBCT models for mesiodistal widths of maxillary right canine, right first molar, left first premolar, left second premolar and left first molar, mandibular right second premolar, left first and premolar, mandibular inter-canine width. Also Statistically significant differences were found between conventional models and digitally scanned models for most mesiodistal teeth widths. However, the mean differences between the measurements were ranging from -0.222 – 0.211 mm which is not clinically significant. For reliability evaluation results showed excellent agreement for the three methods between measurements made at three different time intervals. Conclusion: Digital CBCT models and digitally scanned plaster models used in this study are as accurate as conventional models and can be a good alternative. The three methods used in this study are highly reliable.