Skip to content
ScienceQuest
Biology Visualiser Undergraduate

Full Skeleton Explorer

Free 3D human skeleton: click any of 203 bones to name, isolate or hide it, both sides, with the cartilage and membranes alongside. No sign-in.

Visualiser

Full skeleton, 203 bones in 248 meshes

Checking this device for a stored copy.

Start here

  1. Collapse every band but one The structure list opens closed on a model this size. Open a single band to work through it rather than scrolling 248 rows.
  2. Hide Cartilage and Membranes and ligaments Leaves the 203 bones on their own. The other 45 structures are not bone.
  3. Turn on Fade the rest, then select one bone The only practical way to place a single bone in a whole body. Drag the fade slider to decide how much context you keep.

Bone, cartilage, membranes and ligaments. No coccyx and no auditory ossicles, because the dataset publishes neither, and no teeth. It does file an eyeball and two sets of gums under the skull, which are left out as not skeletal at all. Everything else soft is absent.

  • 203 bones
  • 2362k triangles
  • 12000 KB, on request
  • 0.0125 mm worst error per axis
  • 23 skull and hyoid
  • 24 vertebral column
  • 27 thorax
  • 4 shoulder girdle
  • 60 upper limbs
  • 3 pelvis
  • 62 lower limbs
  • 39 cartilage
  • 6 membranes and ligaments

The model also carries 39 of cartilage and 6 of other tissue, so 248 parts load in total. They are listed and coloured like the rest, just not counted as bone.

203 bones, within 248 meshes, against the 206 a textbook gives, and the gap reconciles exactly. Seven of the 206 are not here, because BodyParts3D publishes no coccyx and none of the six auditory ossicles. Four bones are here that the 206 does not count: the two sesamoids under each big toe. So 206, minus seven, plus four, is 203. The bone meshes come to 203 as well, by a different route: the sternum is three meshes, the manubrium, the body and the xiphoid process, where the 206 counts it as one bone, and each foot’s two sesamoids share a single mesh. The other 45 meshes are 23 intervertebral discs, 16 costal cartilages, 4 interosseous membranes and 2 long plantar ligaments, none of which is bone.

Where this comes from

Segmented from imaging of one adult male body, not drawn by an artist, so the shapes are measurements. Two consequences worth knowing: it is one person’s anatomy rather than an average, and faint stepping on some surfaces is the resolution of the original scan rather than a rendering fault.

Meshes from BodyParts3D, © The Database Center for Life Science, under Creative Commons Attribution 4.0 International. Names and identifiers are the dataset’s own: Foundational Model of Anatomy identifiers and English labels, as its parts list gives them. Converted meshes are shared under the same licence.

Changes made: Meshes were converted from binary STL to an indexed mesh with 16-bit quantised vertex positions, and surface normals were recomputed from the geometry. No vertex was moved other than by that quantisation, whose worst-case error on any one coordinate is reported on the page.

Latin terms for 70 of the 248 structures come from English Wikipedia anatomy infoboxes, under Creative Commons Attribution-ShareAlike 4.0, joined by Wikidata property P1402, Foundational Model of Anatomy identifier. Only a term a source states for that structure is used, never one constructed from a family name, which is why the rest carry no Latin rather than a guess.

All 248 structures

Every structure the model contains, group by group and alphabetically within each. Of these, 203 are bones of this region. Each name opens the viewer with that structure selected, so a link can point at one structure rather than at the whole region.

Skull and hyoid 23

Shoulder girdle 4

Upper limbs 60

Pelvis 3

Lower limbs 62

Cartilage 39

Membranes and ligaments 6

Citing this tool

Last updated . Add the date you accessed it as well, which a citation of a page that can change asks for. If a specific view matters, cite the page address as it stands in your browser instead: it records what is selected and shown.

203 bones, and why not 206

This 3D model of a whole adult human skeleton, from the BodyParts3D dataset, names every bone it holds, 203 of them against the 206 a textbook counts. The familiar figure counts differently rather than counting something else, and the gap closes exactly:

Plus four
The two sesamoid bones under each big toe, which the 206 leaves out. The dataset publishes each foot’s pair as one mesh, two separate pieces side by side under the head of the first metatarsal.
Minus one
There is no coccyx. BodyParts3D publishes the coccygeus muscles and no coccyx bone.
Minus six
No auditory ossicles. The three in each middle ear are not in this dataset either.

206 plus four minus one minus six is 203. The bone meshes come to 203 as well, by a different route: the sternum is here as its three parts, the manubrium, the body and the xiphoid process, where the textbook figure treats it as one bone, and the four sesamoids are two meshes.

Forty-five of the 248 are not bones

The dataset’s skeletal system is not only bone, so they are banded separately rather than counted as bone: 23 intervertebral discs and 16 costal cartilages, which are cartilage, and 4 interosseous membranes and 2 long plantar ligaments, which are neither. Hide those two bands to see the 203 alone.

An earlier version of this page banded the discs with the vertebrae and the cartilages with the ribs, which made it claim 251 bones. It was wrong, and no arithmetic on the page could be made to work.

No published bone is missing, and that is checked

The part list is generated from the dataset’s part-of hierarchy, and a test asserts it matches the leaves under the skeletal system root exactly, in both directions, less three named exclusions. A bone cannot be dropped and one cannot be invented. That test caught two real errors: a first pass banded only 231 parts, because twenty bones are filed by shape rather than by region, and the subtree turned out to contain an eyeball and two sets of gums, because the hierarchy records what a body region contains rather than what it is made of.

The whole human skeleton standing, seen from the front and left, with the skull, spine, rib cage, shoulder girdle, limbs and pelvis each in a different colour.
203 bones plus the cartilage and membranes the dataset files with them, coloured by region, from the frontal bone to the toes. Image © ScienceQuest, CC BY 4.0, rendered from BodyParts3D, © The Database Center for Life Science, CC BY 4.0. Free to reuse with both credits and a link to this page; how to reuse it. Download image

Common questions

How many bones does this model have, and why is it not 206?

203 bones, and the gap from 206 reconciles exactly. Seven of the 206 are not here, because BodyParts3D publishes no coccyx and none of the six auditory ossicles. Four are here that the 206 leaves out: the two sesamoid bones under each big toe. So 206 minus seven plus four is 203. The bone meshes come to 203 as well, by a different route: the sternum is three meshes, the manubrium, the body and the xiphoid process, where the textbook figure treats it as one bone, and each foot’s two sesamoids share one mesh. The model carries 248 structures in total, because 45 of them are not bones: 23 intervertebral discs and 16 costal cartilages are cartilage, and 4 interosseous membranes and 2 long plantar ligaments are neither. Those sit in their own bands rather than being counted as bone.

Is anything missing?

Of the textbook 206, only the coccyx and the six auditory ossicles, which BodyParts3D does not publish; no bone the dataset does publish is missing, and that is checked rather than claimed. The part list is generated from the dataset’s own part-of hierarchy, and a test asserts it matches the leaves under the skeletal system root exactly, in both directions, less three named exclusions. A bone therefore cannot be dropped without failing a test, and one cannot be invented either. Two things that test caught: a first pass banded only 231 parts and looked perfectly healthy, because twenty bones are filed by shape rather than by region, and the subtree also contains an eyeball and two sets of gums, because the dataset records anatomical containment rather than tissue class. Those three are the exclusions.

Why is it so large to download?

Because it is 2,362,198 triangles, which is about 12 megabytes compressed. The ribs and the scapulae alone are about two fifths of it: long thin curved shells are the worst case for a triangle budget, since a surface enclosing almost no volume has nothing to amortise its own area against. Nothing downloads until you press the button and the figure is printed on it, so this costs nothing unless you want it, and once you have it the model is kept on your device and opens without asking next time. If you only need one area, the individual regions are a fraction of the size.

Will this run on a phone?

It may not, and as one of the heavier models here it is a real question. Two and a third million triangles needs roughly 45 megabytes of graphics memory before anything is drawn, which is comfortable on a laptop and near the limit on an older phone. If it fails the viewer falls back to the structure list rather than showing an error, so you still get every name, and the smaller regions are a better choice on a phone regardless: a single limb is under a megabyte and the skull only a little over one, and both turn smoothly.