Cephalometric Landmarks: Definitions and Common Errors
Every cephalometric measurement is only as good as the landmarks behind it. A Nasion placed too high, a Gonion picked on the wrong contour, an incisor apex stopped halfway, and SNA, the mandibular plane or the incisor inclination change without any change in the patient. This article is a reference for the landmarks themselves: definitions by region, the pitfalls taught in university orthodontic programs, how an error spreads to several measurements, and what research says about reproducibility. The tracing procedure itself, from checking the radiograph to interpreting the numbers, is covered step by step in our cephalometric tracing guide.
Anatomical, constructed and bilateral landmarks
Landmarks fall into two families. Anatomical landmarks are identified directly on the radiograph (Sella, Menton, an incisal edge). Constructed landmarks are obtained from lines (Gonion by a bisector, the projections of points onto S-N). Two definitions of the same landmark can coexist, and they do not always give the same position.
A second distinction matters even more. Midline structures (Sella, Nasion, the symphysis, the incisors) project as a single image. Bilateral structures (ear canals, orbits, mandibular angles and borders, molars) project twice, offset by magnification, head rotation or a true asymmetry. Most of the hardest landmarks are bilateral.
In the tables below, landmarks in bold are the 15 landmarks of our free cephalometric analysis tool, which are enough for the Steiner and Tweed analyses. The others are used by other measurements or analyses.
Cranial base, ear and orbit
| Landmark | Definition | Used for | Pitfalls |
|---|---|---|---|
| S (Sella) | Center of the sella turcica | S-N line, SNA, SNB, SN-GoGn | Center estimated by eye: see the whole contour first. Look at the shape of the sella before measuring (an abnormal sella calls for a specialist opinion) |
| N (Nasion) | Most anterior point of the frontonasal suture | S-N, N-A, N-B: SNA, SNB, ANB, incisor measurements | Faint suture; one error affects every Steiner measurement at once |
| Po (Porion) | Highest point of the external auditory meatus | Frankfort plane (FMA, FMIA) | Doubled, low contrast; the canal is an oval of about 8 to 10 mm, its long axis about 45° to Frankfort; do not confuse it with the ear rods of the cephalostat |
| Or (Orbitale) | Lowest point of the orbital rim | Frankfort plane | Two superimposed orbits: apply one convention, often the midpoint |
| Ba (Basion) | Anterior margin of the foramen magnum | Other analyses (Ba-N, N-S-Ba) | Crowded area; courses place the tip of the odontoid process 4 to 6 mm below Basion |
| E | Perpendicular projection onto S-N of the tangent to the posterior border of the condyle | Steiner’s SE (22 mm ± 2) | Depends on the condyle outline, often doubled |
| L | Perpendicular projection of Pogonion onto S-N | Steiner’s SL (51 mm ± 2) | Inherits any error on Pogonion and on S-N |
The maxilla
| Landmark | Definition | Used for | Pitfalls |
|---|---|---|---|
| A (subspinale) | Deepest (most posterior) point of the anterior alveolar concavity of the maxilla, between the anterior nasal spine and prosthion | SNA, ANB, U1-NA, Wits | Low contrast; a periosteal point that remodels with incisor movement |
| ANS | Anterior end of the bony palate (anterior nasal spine) | Palatal plane, lower anterior face height | Among the least reliable landmarks in reliability studies (see below) |
| PNS | Posterior end of the bony palate | Palatal plane | Low-contrast area: trace the whole palate outline before choosing the point |
| Ptm | Apex of the pterygomaxillary fissure | Other analyses | Bilateral fissures, often doubled |
The mandible
| Landmark | Definition | Used for | Pitfalls |
|---|---|---|---|
| B (supramentale) | Deepest point of the anterior concavity of the symphysis | SNB, ANB, L1-NB, Wits | Influenced by the position of the lower incisor |
| D | Center of the body of the symphysis | Steiner’s SND (76° ± 2) | Not influenced by the incisor, unlike B; but it is an estimated center |
| Pog (Pogonion) | Most anterior point of the symphysis | Pog-NB, Holdaway ratio, A-Pog | Faint chin outline; constructed points depend on it |
| Pm | Suprapogonion | Other analyses | Depends on how the chin curvature is read |
| Gn (Gnathion) | Point on the outline of the symphysis between Pogonion and Menton | Steiner’s mandibular plane (Go-Gn) | Two constructions are taught (see below) |
| Me (Menton) | Lowest point of the symphysis | Mandibular plane, lower anterior face height | Blurred lower border in some images |
| Go (Gonion) | Point at the mandibular angle | Go-Gn, mandibular plane | Double contour; two definitions (see below) |
| Ar (Articulare) | Crossing of the posterior border of the ramus and the clivus | Posterior facial height (Tweed) | The crossing of two shadows, not a bony point: it moves when either outline is doubled |
| Xi | Center of the ramus | Other analyses | Constructed from the ramus outline |
Gnathion and Gonion: one name, two definitions
Gnathion is either constructed (the bisector of the angle between N-Pog and the mandibular plane crosses the symphysis, as in Steiner’s course) or simply placed midway between Pogonion and Menton, as in another handout. Our free tool uses the descriptive version: a point on the symphysis outline between Pogonion and Menton.
Gonion is constructed in courses: the bisector of the angle between the tangent to the posterior border of the ramus and the mandibular plane, carried to the angle. Our tool uses the most posterior and inferior point of the angle. The two can differ slightly: compare measurements obtained with the same definition.
The double mandibular contour
When the two halves of the mandible do not superimpose, the courses (after Bassigny) give a rule:
- if the two contours cross at the angle, trace the most posterior border for the ramus and the highest border for the body;
- if they are offset without crossing, trace the midline between them and note a possible sign of asymmetry.
The literature adds that a double contour of the rami without any intersection has been described as pathognomonic of a mandibular laterognathia (Choquin). A doubled image therefore deserves a look at head positioning, then at the face; our article on asymmetries in orthodontics (in French) covers the work-up.
The teeth
| Landmark | Definition | Used for | Pitfalls |
|---|---|---|---|
| UIE / UIA | Incisal edge, then root apex, of the upper central incisor | U1-NA (°), interincisal angle | The apex is hidden by neighboring teeth and bone: follow the root to its end |
| LIE / LIA | Incisal edge, then root apex, of the lower central incisor | L1-NB (°), IMPA, FMIA, interincisal angle | Choose the most labially positioned incisor; in the mixed dentition, do not confuse a tooth with a germ |
| UIL / LIL | Most labial point of the upper and lower incisor crowns | U1-NA and L1-NB in mm | Optional in our tool; need calibration |
| First molars | Middle of the overlap of the first molar cusps | Occlusal plane (Occ/SN, Occ/FH, Wits) | Two molars per arch superimposed; delicate in the mixed dentition |
The occlusal plane runs from the middle of the first molar overlap to the middle of the incisor overlap. Steiner relates it to S-N (14°), Tweed to Frankfort (10°, 8 to 12°): never carry one value into the other analysis.
Soft-tissue landmarks
| Landmark or measurement | Definition | Used for |
|---|---|---|
| Soft-tissue pogonion | Most anterior point of the soft-tissue chin | S line, Z line, E line |
| Most prominent lip | The more protrusive of the two lips | Merrifield’s Z line (angle Z, 75° ± 5) |
| Nasal point of the S line | Middle of the S-shaped curve at the base of the nose | Steiner’s S line (lips should just touch it) |
| Upper lip thickness (UL) | From the most prominent point of the upper lip to the labial surface of the upper central incisor | Tweed-Merrifield, compared with TC |
| Total chin (TC) | From soft-tissue pogonion to its projection on N-B | Expected to be at least equal to UL |
Ricketts’ E line is also quoted (upper lip about 1 mm behind it, lower lip on it). Soft-tissue points need contrast on the profile and lips at rest: tense lips change every soft-tissue value.
How landmark error propagates to measurements
Only the perpendicular part counts
An error matters only in its component perpendicular to the line the landmark defines: moving point A along N-A barely changes SNA, moving it forward or back changes it directly. A study of 1,116 tracings by three raters found exactly this: the variance orthogonal to the defining planes was relevant, the variance parallel to them was not, and each landmark showed a characteristic "error envelope" (Grogger et al., 2018, Int J Oral Maxillofac Surg, PubMed). In that study, the horizontal coordinate of point A was the main source of error for SNA, ANB and the Wits appraisal, and the vertical coordinate of Nasion dominated the error of SNB. The sample was made of surgical patients with dentofacial deformities.
Distance from the vertex
The angular error is roughly the perpendicular error divided by the distance from the vertex of the angle (in radians). By simple trigonometry, 1 mm at 60 mm from the vertex shifts a line by about 1°, while 1 mm on a 20 mm segment shifts it by about 3°. This is why short lines, such as an incisor long axis, are so sensitive to an apex placed a little too short.
One landmark, several measurements
In our free tool, each landmark feeds the following measurements:
| Landmark | Measurements affected |
|---|---|
| N | SNA, SNB, ANB, SN-GoGn, U1-NA and L1-NB (degrees and mm) |
| A / B | SNA and ANB, U1-NA / SNB and ANB, L1-NB |
| Po, Or | FMA, FMIA |
| Go | SN-GoGn, FMA, IMPA, FMIA |
| Gn | SN-GoGn |
| Me | FMA, IMPA, FMIA |
| Incisal edges and apices | Incisor angles, interincisal angle, IMPA, FMIA |
| Labial crown points | U1-NA and L1-NB in mm (with calibration) |
A misplaced Nasion distorts the whole Steiner side at once, a misplaced Gonion both divergence measurements. One caution: in our tool, FMIA is derived as 180 − FMA − IMPA, so the Tweed triangle always closes. The 180° check helps when the three angles are measured separately, as in manual tracing; it cannot catch an error on Frankfort.
Distances add a further source of error: magnification (about 7% with the source at 1.5 m, according to courses) and calibration. Without calibration, millimeter values cannot be compared with norms.
How reproducible are cephalometric landmarks?
- Some landmarks are much less reliable than others. On 20 digital cephalograms, the lower incisor edge was the most consistently identified landmark, while condylion, gnathion, orbitale and the anterior nasal spine were the least reliable; agreement within one observer was excellent, but only good between observers (Durão et al., 2015, Imaging Sci Dent, PubMed). The sample is small and the reference was the mean of the observers.
- Porion and pogonion divide observers. On 50 patients traced three times by two observers, the only significant differences between observers concerned porion, pogonion and a ramus point; film and monitor gave the same results (Yu et al., 2008, Am J Orthod Dentofacial Orthop, PubMed).
- Training matters. In 277 analyses by 161 dental students, compared with a reference set by six orthodontists, the mean landmark deviation was 2.05 mm, and 68.6% of landmarks fell within the 2 mm threshold often described as clinically acceptable, with large differences between landmarks (Kanemeier et al., 2024, Head Face Med, PubMed).
- Constructed is not necessarily worse. In an older study (12 orthodontists, 4 radiographs), constructed landmarks were as reliable as directly identified ones (Savage et al., 1987, Am J Orthod Dentofacial Orthop, PubMed).
- Midline beats bilateral. In CBCT studies, midsagittal and dental landmarks were the most reliable, and condyle, porion and orbitale the least (Lisboa et al., 2015, J Appl Oral Sci, PubMed). The data are three-dimensional, but they confirm the bilateral problem seen in 2D.
A broader caution: a systematic review found the evidence on the usefulness of the lateral cephalogram for treatment planning limited and contradictory (Durão et al., 2013, Prog Orthod, PubMed). Landmarks support a diagnosis; they do not make it.
Manual tracing and AI-placed landmarks
Our free tool is manual: you place each landmark yourself, which is the best way to learn them.
Software can also place landmarks automatically. It tends to struggle where the human eye hesitates: Porion, Orbitale, Gonion, incisor apices, points A and B. Every automatically placed landmark must be checked and validated before the measurements are used. Our article on artificial intelligence in cephalometrics (in French) explains how these models work and where they fail.
In practice: a checklist
- Read the radiograph before placing landmarks: sella, vertebrae, double contours.
- Fix one convention for bilateral landmarks and one definition of Gnathion and Gonion, and keep them for all of a patient’s tracings.
- Check Nasion, A and B first: they drive SNA, SNB and ANB.
- Follow each incisor root to its apex, and calibrate before reading any distance.
- If a measurement contradicts the clinical examination, recheck the landmarks before concluding.
Frequently asked questions
How many cephalometric landmarks are there?
There is no fixed number: each analysis uses its own set. Our free tool uses 15 landmarks, enough for Steiner and Tweed. Public research datasets annotate 19 (ISBI 2015) or 29 landmarks including soft tissue (Aariz). Ricketts, Delaire or airway analyses add more.
Which cephalometric landmarks are the least reliable?
Bilateral and low-contrast landmarks: Porion, Orbitale, Gonion, condylion, plus Gnathion depending on its definition, the anterior nasal spine, point A and the incisor apices. Midline landmarks with sharp outlines, such as the incisal edges and Menton, are the most reproducible.
How accurate does landmark placement need to be?
Studies often treat 2 mm as the clinically acceptable threshold. What matters most is the direction of the error: an error perpendicular to a reference line changes the measurement, an error along it barely does.
With GS Ortho
In GS Ortho, AI places 29 landmarks on the lateral cephalogram, soft-tissue profile included, and the Steiner and Tweed analyses are calculated straight away. You check each landmark, correct it if needed, then validate: the decision stays yours. The tracing goes into the patient’s record, the measurements feed into the diagnostic studio, and an orthodontist’s opinion can be requested in one click, straight from the case. The GS Ortho interface is currently available in French.