Purpose: The purpose of this study was to report the reproducibility of intra-observer and inter-observer consistency of cephalometric measurements using three-dimensional (3D) computed tomography (CT), and the degree of difference of the cephalometric measurements. Materials and Methods: CT images of 16 adult patients with normal class I occlusion were sent to personal computer and reconstructed into 3D images using V-Works 3.5^TM(Cybermed Inc., Seoul, Korea). With the internal program of V-Works 3.5^TM, 12 landmarks on regular cephalograms were transformed into 21 analytic categories and measured by 2 observers and in addition, one of the observers repeated their measurements. Intra-observer difference was calculated using paired t-test, and inter-observer by two sample test. Results: There were significant differences in the intra-observer measurements (p < 0.05) in four of the categories which included ANS-Me, ANS-PNS, Cdl-Go (Lt), GoL-GoR, but with the exception of Cdl-Go (Lt), ZmL-ZmR, Zyo-Zyo, the average differences were within 2 mm of each other. The inter-observer observations also showed significant differences in the measurements of the ZmL-ZmR and Zyo-Zyo categories (p < 0.05). With the exception of the Cdl-Me (Rt), ZmL-ZmR, Zyo-Zyo categories, the average differences between the two observers were within 2 mm, but the ZmL-ZmR and Zyo-Zyo values differed greatly with values of 8.10 and 19.8 mm respectively. Conclusion: In general, 3D CT images showed greater accuracy and reproducibility, with the exception of suture areas such as Zm and Zyo, than regular cephalograms in orthodontic measurements, showing differences of less than 2 mm, therefore 3D CT images can be useful in cephalometric measurements and treatment planning.