Dinosaur classification: the family tree explained

Dinosaur classification is a hypothesis about ancestry, not a set of drawers chosen for convenience. Palaeontologists compare hundreds or thousands of anatomical features, record which form occurs in each species, and use those observations to find family trees that require the fewest unsupported evolutionary changes. A new fossil or a revised interpretation of one bone can therefore move a species – and sometimes alter the deepest branches of the tree.

Three major lineages are consistently recognised: Theropoda, Sauropodomorpha and Ornithischia. Their internal branches are supported by large bodies of evidence. The difficult problem lies close to their common origin in the Late Triassic: analyses have recovered every possible pairing of the three. A useful classification must show this uncertainty instead of presenting one attractive diagram as settled fact.

Classification at a glance

What makes an animal a dinosaur?

A dinosaur is not simply any large extinct reptile. Pterosaurs, marine reptiles such as plesiosaurs, and synapsids such as Dimetrodon fall outside Dinosauria. Dinosaurs belong to Archosauria, the larger group that also includes crocodilians and pterosaurs. Within the bird-line archosaurs, their closest known relatives include silesaurids, a varied assemblage of generally small Triassic animals whose exact position is itself debated.

No single bone works as an infallible membership badge. Researchers diagnose Dinosauria with a combination of features, many concentrated around the pelvis, hind limb and shoulder. Examples include a perforated or partly open hip socket in many early members, an enlarged attachment on the upper arm, and changes in the ankle and thigh that brought the limbs beneath the body. Some features also occur in close relatives, while others are absent or transformed in specialised dinosaurs. Membership is decided from the whole anatomical pattern and position on a tested tree.

How a cladogram is built

A phylogenetic analysis begins with taxa and characters. A character might record whether a particular bone is straight or curved, whether two bones contact, or whether a tooth has a specific shape. Researchers divide each character into explicitly defined states and score them from specimens, photographs and published descriptions. A computer then searches the enormous number of possible trees for those that best fit the distribution of states under the chosen method.

The branches in the result represent nested hypotheses of common ancestry. A node is not automatically a named rank such as order or family, and vertical position does not mean that one group is “more evolved”. All terminal species have travelled from their shared ancestor to their own time. Rotating branches around a node changes the drawing but not the relationships.

Support values test how consistently the dataset recovers a branch. They do not turn missing fossils into evidence. Closely related early dinosaurs often share primitive anatomy, and many specimens are incomplete or distorted. Different teams may define characters differently, score uncertain structures differently, choose different species, or use different models. Sensitivity analyses reveal whether a conclusion survives those choices.

The traditional split: Saurischia and Ornithischia

In 1887, Harry Govier Seeley divided dinosaurs mainly by pelvic construction. Saurischia, the “lizard-hipped” group, traditionally combines theropods with sauropodomorphs. Ornithischia, the “bird-hipped” group, includes stegosaurs, ankylosaurs, ornithopods, pachycephalosaurs and ceratopsians. This broad scheme dominated dinosaur classification for more than a century and remains one of the plausible arrangements recovered by modern analyses.

The names are historically useful but easy to misread. Birds did not arise from ornithischians. They evolved within Theropoda. In many advanced theropods the pubis also rotated backwards, but that happened independently from the broadly similar condition that inspired the name Ornithischia.

The 2017 Ornithoscelida challenge

A large 2017 analysis grouped Theropoda with Ornithischia in a clade called Ornithoscelida, leaving Sauropodomorpha on the other major branch. That result revived a nineteenth-century name and drew attention to anatomical similarities between early theropods and early ornithischians. It also implied that several familiar character changes occurred in a different order than the traditional tree suggested.

A rapid reanalysis by another international team revised many character scores and recovered traditional Saurischia, although the competing arrangements were statistically difficult to separate. The original authors replied with additional analyses that continued to favour Ornithoscelida. The exchange was valuable precisely because the disagreement could be traced to explicit observations and modelling decisions rather than authority.

A third arrangement and the silesaurid problem

Other datasets have paired Sauropodomorpha with Ornithischia, an arrangement sometimes labelled Ornithischiformes. This possibility becomes especially important when silesaurids are placed on the ornithischian stem or within the earliest part of Ornithischia. Those placements can help fill an otherwise conspicuous gap: secure theropods and sauropodomorphs occur early in the Late Triassic, while unmistakable ornithischians appear later and are initially rare.

Silesaurids combine dinosaur-like features with anatomies that do not fit neatly inside the classic three branches. Some have beak-like lower jaws and teeth compatible with herbivory, but their hips and ankles preserve a mosaic of states. Analyses can place them just outside Dinosauria, as successive relatives of dinosaurs, or close to the origin of Ornithischia. Their position affects both the definition of Dinosauria and the inferred body plan of its earliest members.

A broad 2024 review found no emerging consensus at the base of the dinosaur tree. Recent studies collectively recover all three possible pairings among Theropoda, Sauropodomorpha and Ornithischia. That does not make classification arbitrary: the uncertainty is concentrated in short early branches and incomplete Triassic material, while many later groups remain strongly supported.

Theropoda: the branch that includes birds

Theropods began as mostly bipedal dinosaurs with grasping hands and blade-like teeth, but the lineage diversified far beyond large predators. Ceratosaurs, megalosauroids, allosauroids, tyrannosauroids and coelurosaurs represent only some of its major branches. Our profiles of Tyrannosaurus rex and Velociraptor occupy very different positions within Coelurosauria and should not be treated as interchangeable “raptors”.

Feathers or simpler filamentous coverings occur across several theropod groups. Within Maniraptora, changes to the wrist, shoulder, tail and feathers preceded powered flight. Birds are avialan theropods and therefore living dinosaurs. The phrase “non-avian dinosaur” is used when writers mean every extinct dinosaur lineage outside the bird crown group.

Sauropodomorpha: from light bipeds to giants

Early sauropodomorphs were not simply miniature sauropods. Many were relatively light, sometimes facultatively bipedal animals with small heads and lengthening necks. Across the lineage, changes in limb proportions, vertebral construction, feeding reach and growth eventually produced the column-limbed sauropods. Their air-filled vertebrae reduced skeletal mass, but body mass still has to be estimated from limb dimensions or three-dimensional volume. The assumptions behind those methods are demonstrated in our dinosaur size comparison.

The Late Triassic sauropodomorph Mbiresaurus raathi from Zimbabwe is important because it belongs to one of Africa’s oldest secure dinosaur assemblages. Its occurrence alongside other early dinosaur groups resembles assemblages at comparable southern palaeolatitudes, supporting the idea that climate belts constrained the first dinosaur dispersals across Pangaea.

Ornithischia: repeated experiments in herbivory

Ornithischians share a distinctive predentary bone at the front of the lower jaw and other cranial and pelvic features. Their major lineages include armoured thyreophorans, ornithopods, dome-headed pachycephalosaurs and horned ceratopsians. Different groups independently evolved complex chewing systems, body armour, display structures and large body size.

Names such as “duck-billed dinosaur” or “armoured dinosaur” describe appearance, not formal rank. Hadrosaurids sit inside Ornithopoda; ankylosaurs and stegosaurs are branches of Thyreophora. A family tree keeps those nested relationships visible and prevents a list of familiar body types from being mistaken for equal taxonomic units.

Why the tree changes without collapsing

Changing the deepest dinosaur split does not erase everything known about dinosaurs. The three principal lineages remain recognisable, birds remain theropods, and most well-supported subgroups retain their membership. What changes is the sequence in which a small set of early branches diverged and, consequently, how often researchers infer that particular traits evolved.

Future progress is most likely to come from better Triassic fossils, detailed redescription of existing museum specimens, three-dimensional imaging, and datasets that publish their scoring decisions openly. Computational methods can explore uncertainty, but they cannot compensate for a misidentified or missing bone. The tree is therefore a testable summary of current evidence, not a finished diagram.

How to read classifications in this atlas

Each profile in the dinosaur encyclopedia uses the narrowest stable placement supported by the available material. A secure family or subfamily is named when diagnostic anatomy justifies it. Fragmentary species are not forced into a precise branch merely to fill a table. When competing analyses matter to the animal’s interpretation, the alternatives are stated separately.

Scientific names can also change without the fossil changing. A species may be transferred to a different genus, two names may be recognised as synonyms, or a historically combined assemblage may be split. The original name attached to the type specimen remains part of the nomenclatural record; classification asks where that named species belongs on the tree.