OBJECTIVE To compare measurements from human skulls and their images from cephalometric radiographs (CR) and computed tomography (CT) scanograms, in order to gauge the potential clinical use of the latter. METHODS Based on specific inclusion criteria, including stable centric occlusion, 13 adult skulls were selected from a larger collection. The mandible was taped to the maxilla after securing the occlusion of teeth and condylar seating in the glenoid fossa. Lateral and posteroanterior cephalographs and CT 'scout views' were taken of each skull by standardized methods. Landmarks were identified on skulls and images. METHODS Linear measurements were made on all three records; angular measurements only on CR and CT images. Intraclass correlation coefficients (r) were calculated to assess similarity among records. Paired t-tests were used to compare differences between mean measurements. RESULTS No statistically significant differences were noted between mean angular values on CR and CT views (P > 0.05). The highest correlations were observed for several vertical midline distances between CT and direct skull measures: 0.82 < r < 0.995-greatest for nasion-menton. For sagittal distances, the highest correlation was between the direct measure of condylion-pogonion and its CR image (r= 0.73). Correlations between CR and skull transverse measures were higher (0.46 < r < 0.80) than the corresponding skull vs. CT measures (0.06 < r < 0.38). CT and CR images are 2D slices and projections, respectively, of 3D structures. Vertical CT and skull measures correspond because the CT projection reflects a 1:1 ratio in the midsagittal plane; CT projected lateral images are smaller than the skull measures. The CR image reflects a distortion (approximately 8%) that brings Co-Pg closer to its anatomic distance, inadvertently contributing to better clinical planning, particularly in orthognathic surgery. The pattern of distortion of PA images was in opposite directions for CR and CT views. CONCLUSIONS Cephalograms and CT scanograms are close in depicting angular relations of structures, but they differ in the accuracy of imaging linear measurements, because the location and size of an object within the imaged 3D structure varies with both records. Logistic and economic considerations favor the use of cephalographs.