This study was undertaken to quantify the reproducibility of 14 landmarks used in cephalometric analysis and to examine inter- and intra-observer differences. A special study was made of the point Apex inferior. A hundred headplates (50 boys and 50 girls) taken at random from the 12-year-old group of the Nittedal collection in the Orthodontic Department of the University of Oslo were used for the investigation. Landmarks were registered on two occasions by each of two observers and recorded on a co-ordinate system with the X-axis approximately parallel to the Frankfurt plane. Scattergrams to show inter- and intra-observer variability showed that in general the precision was good though there were variations between different points. The scattergrams reflected the radiographic anatomic characteristics of each landmark and the difficulties in localization. To avoid single gross errors the clinician is advised to remeasure each headplate after an interval. It should be noted that there are inter-observer differences in the interpretation of the definition and the localization of landmarks. Apex inferior was not located with confidence by the observers in 75% of the cases.