OBJECTIVE To identify anatomic structures in three dimensions and examine their reliability to be used as landmarks in a three-dimensional coordinate cephalometric analysis, using cone-beam computerized tomography (CBCT). METHODS Thirty CBCT images were randomly selected for landmark location. Forty-two anatomic landmarks, which are not included in the traditional cephalometric landmarks, were chosen based on radiographic characteristics that make them pragmatic to mark in the CBCT image slices. The principal investigator marked the full set of landmarks on the software by navigating in the X, Y, and Z axes for every image three times, with each measurement trial being at least 1 week apart. One other investigator also located the landmarks once for each image for reliability purposes. Intraclass correlation coefficients (ICCs) were used to analyze the mean differences in landmark location in all axes. RESULTS Intra- and interexaminer reliability for x, y, and z coordinates for all landmarks had ICC greater than 0.95 with confidence interval of 0.88-0.99. Mean measurement differences found were <1.4 mm for all landmarks in all three coordinates. Mean measurement error differences obtained in the principal investigator's trials were primarily <0.5 mm. CONCLUSIONS The most reliable and reproducible landmarks tested for use in CBCT are mental foramina, infraorbital foramina, inferior hamulus, dens axis, foramina transversarium of atlas, medial and lateral condyles of the mandible, superior clinoid processes, and mid-clinoid.