What did dinosaurs eat?

Dinosaurs occupied food webs for more than 160 million years. They included hunters of large prey, fish catchers, leaf and seed eaters, animals that cropped woody vegetation, and forms able to switch among resources. The question cannot be reduced to carnivore versus herbivore. Evidence has to be assembled for each species, age and environment.

The strongest answers come from food preserved inside a body, fossilised faeces and tooth marks on bone. Teeth and jaws show what could be gripped, cut or ground. Wear records repeated contacts. Biomechanical models test whether a proposed action fits the structure. No method supplies a complete menu, but converging clues can turn an attractive guess into a testable conclusion.

Which evidence deserves the most confidence?

Every clue has a different limit. Stomach contents record a particular meal but may be unusual. A tooth identifies a tool but seldom names a food. A coprolite contains real digested matter, yet may lie far from its maker. A bite mark connects a tooth to bone without always distinguishing a kill from scavenging.

Researchers therefore ask whether material really lay inside the body or digestive tract, whether it can be associated with a particular animal, and whether anatomy, wear and other specimens support the same conclusion. A striking inclusion is strongest when its position and taphonomy rule out accidental mixing after death.

The time represented by each clue also differs. Gut contents may capture hours, microwear days or weeks, and tooth shape an evolutionary adaptation maintained across generations. Stable anatomy describes capacity, while a last meal describes behaviour on one occasion. Treating those scales as identical creates false certainty. A convincing reconstruction states which time span and which individual each observation actually covers.

A tooth is a tool, not a ready-made menu

A recurved crown with serrated edges penetrates and draws through tissue. A long conical crown resists bending while holding a slippery object. Low leaf-shaped crowns with coarse marginal denticles crop vegetation. Hundreds of replacement teeth packed into a battery form a continuous working surface. These patterns expose mechanical tasks rather than a shopping list.

Baryonyx had conical teeth suited to fish capture, yet remains found with its skeleton included both fish and a young herbivorous dinosaur. A leaf-shaped crown indicates plant processing without choosing among fern, conifer, fruit or flowering leaf. Even similar diets produced very different dental systems.

How carnivorous dinosaurs fed

Most large Mesozoic predatory dinosaurs were theropods, but their teeth were not interchangeable. Many large theropods carried recurved crowns with serrations. Complex folds of dental tissue strengthened each denticle. The arrangement could puncture flesh and extend a cut through repeated bites.

Tyrannosaurus rex had unusually thick crowns able to tolerate heavy loading. Bite traces and shattered bone show that it did not stop at soft tissue. It could remove and swallow pieces of bone, while digestion dissolved part of the mineral component. Bone was not necessarily the target. Robust teeth allowed deeper feeding on a carcass and survived incidental contact with hard tissue.

Spinosaurids combined narrow snouts with conical teeth. Acid-etched fish scales and teeth occurred in the abdominal area of the Baryonyx holotype, together with bones of a young iguanodontian. That directly confirms fish consumption while rejecting an overly narrow label of exclusive fish eater. Available prey could vary.

Tooth replacement kept these systems functional. New crowns developed inside the jaw and moved into place after loss or breakage. Replacement rate, crown thickness and serration geometry therefore belong to the feeding apparatus just as much as bite force. A predator that regularly contacted bone faced a different balance between sharpness, strength and replacement from one that mostly held small slippery prey.

Herbivores solved the problem in several ways

Plants resist processing through fibres, silica and high water content. Some dinosaurs reduced food in the mouth, whereas others stripped it quickly and relied on long fermentation in a large digestive tract. The absence of a chewing battery does not make a herbivore inefficient.

Sauropod front teeth mainly grasped or cropped vegetation. Narrow Diplodocus crowns were concentrated at the jaw front, while broader spoon-shaped Camarasaurus teeth tolerated a different pattern of contact. Growth lines indicate replacement at roughly 35 days in Diplodocus and about 62 days in Camarasaurus. Rapid renewal compensated for wear without complex chewing.

Hadrosaurs used numerous crowns at once in dental batteries. Enamel, dentine and other tissues wore at different rates, maintaining a rough surface. Ceratopsian batteries formed vertical shearing systems. The teeth of Triceratops also combined tissues with different properties, producing a self-sharpening edge.

Muzzle breadth, neck reach and body height further divided plant resources. A narrow cropping mouth could select branches differently from a broad bill, while a long-necked sauropod could feed over a larger vertical range without moving the whole body. These possibilities still need environmental evidence. Pollen, leaves and wood from the formation define what was available, but presence in a habitat does not prove consumption by every herbivore living there.

Wear records jaw movement and recent food

Large facets develop where teeth meet repeatedly, and their angle records the direction of load. Under a microscope, scratches and pits mark contact with food, adjacent teeth and particles of sediment. Their orientation can reconstruct the working path of a jaw.

Edmontosaurus teeth preserve several consistent scratch directions. They show that the rows did more than close vertically, although microwear alone does not settle which skull movements created every phase. Joint geometry, muscle reconstruction and load models remain necessary. Surface roughness may distinguish food properties, but translating “many pits” directly into a plant name is unsafe. Grit and feeding height can change the signal. The dedicated guide to dinosaur tooth wear separates measured marks from broader dietary inference.

Gut contents preserve an actual meal

Material in the abdominal region counts as direct evidence only after taphonomic testing. Bones or plants should lie inside the body outline, differ from surrounding sediment and ideally show digestive alteration. Otherwise water or burial disturbance may have introduced them.

A juvenile Gorgosaurus from Canada's Dinosaur Park Formation preserved hind limbs of two year-old Citipes in its abdominal cavity. Different digestive states indicate two feeding episodes. The young tyrannosaurid selected small prey and apparently consumed the fleshy legs. The case supports an age-related shift in niche: agile juveniles and massive adults could target different sizes of animal.

Sinocalliopteryx and Microraptor specimens preserved birds and other small vertebrates in their digestive regions. They identify prey and aspects of swallowing, but a single last meal does not turn a species into a narrow specialist.

The last meals of herbivores

The exceptionally preserved nodosaur Borealopelta markmitchelli contained a compact plant mass. Microscopy found mainly fern leaves, smaller amounts of cycad and conifer material, wood fragments and a little charcoal. Comparison with the surrounding flora suggests selection rather than indiscriminate swallowing. It remains one seasonal snapshot from one animal.

In 2025 researchers described the first convincing sauropod gut-content sample, from a juvenile Diamantinasaurus matildae in Australia's Winton Formation. It contained conifer shoots, flowering-plant leaves and reproductive parts of seed ferns. The plants were weakly fragmented, consistent with a broadly feeding animal that cropped at several heights and shifted most processing into the gut. One specimen cannot define every Jurassic and Cretaceous sauropod, whose skulls, muzzles and teeth reveal resource partitioning within the group.

Coprolites record the digested portion

A coprolite is fossilised faecal matter. It may retain bone splinters, scales, wood tissue, leaf cuticle, pollen, spores and phytoliths. These inclusions passed through a digestive system, and their breakdown can reveal digestive intensity.

A large Late Cretaceous Canadian coprolite contains abundant crushed bone and is attributed by size, age and composition to a large theropod, probably a tyrannosaurid. Indian Late Cretaceous coprolites preserve grass phytoliths alongside conifers, palms and other plants. They show that grasses entered some Mesozoic food chains, not that modern open grasslands already existed.

The main difficulty is authorship. An isolated coprolite rarely remains inside a skeleton. Shape, size and contents narrow the candidates without always identifying one genus. Our chapter on coprolites and gut contents treats food as direct evidence and the identity of the producer more cautiously.

Bite marks show feeding, not a complete menu

A groove, puncture or broken bone edge may match a predator's tooth size and shape. A healed injury proves that the victim survived the encounter. Unhealed marks record contact near death without showing whether the predator killed the animal or arrived at an existing carcass.

Feeding traces are unusually common at Colorado's Mygatt-Moore Quarry. Their distribution shows theropods exploiting low-value parts, including bone. That may reflect intensive use of available resources, but cannot become a precise famine calendar or proof of permanent scavenging. Hunting and scavenging are compatible behaviours in a large terrestrial predator.

Did sauropods need stomach stones?

Older reconstructions often gave sauropods a bird-like gastric mill in which swallowed stones ground plants. Polished stones occur near some skeletons, and dinosaurs sometimes swallowed them. The difficulty is demonstrating a large, permanent mill.

The mass of suspected gastroliths is too low relative to sauropod body mass for an efficient avian analogue. Stones could have been swallowed occasionally, helped mix food, or not belonged to the skeleton at all. Microbes and prolonged retention in a large gut probably performed most plant breakdown. Soft organs rarely fossilise, so the exact stomach shape and food passage rate remain models.

Were any dinosaurs omnivorous?

Yes, although omnivory is harder to establish than the label suggests. Teeth able to process several materials merely permit a mixed diet. Small theropods may have combined plants, insects, eggs and small vertebrates, but direct contents are absent for most species.

Quantitative comparison among coelurosaurs shows herbivory was more widespread than the old equation of theropod with meat eater allowed. Toothless beaks, enlarged gut regions and particular crown shapes evolved independently. Therizinosaurs and many oviraptorosaurs fit plant-rich or mixed diets, with differences among species. “Omnivore” should describe a range of possibilities rather than hide uncertainty about proportions.

Diet changed with age, place and season

A species need not occupy one food niche for life. The juvenile Gorgosaurus with small prey shows how growth changed available targets. Tyrannosaurid skulls deepened, teeth thickened and bodies lost juvenile slenderness with age, potentially reducing competition between young and adults.

Plant food also varied by season and region. Ferns, conifers, cycads and flowering plants were unevenly distributed, while feeding height depended on body size. One individual's stomach records a place and brief interval. Several worn teeth, multiple coprolites and plants from the same layer yield a more representative picture.

How a diet is reconstructed

Work begins with crown shape, serrations, roots, replacement rate and tooth arrangement. Researchers then measure skull joints and possible muscle forces. Macrowear is compared with tooth-row movement, microwear with recent contact, and direct food remains with their position, chemistry and burial history.

Living and extinct relatives provide comparisons, not exact replicas. A crocodilian helps explain puncture and grip, a bird may inform gastrolith function, and a large herbivorous mammal offers a model of fermentation. A reliable conclusion remains proportional to the evidence. “This crown tolerated tissue slicing” is stronger than “this species hunted only one prey”. “Several plant groups occur in this gut” is more accurate than “all sauropods ate the same food”.

Three-dimensional models add another check. A digital skull can be loaded in alternative muscle arrangements to find where stress concentrates, but the result depends on estimated soft tissues, material properties and boundary conditions. Models are most persuasive when observed tooth damage, joint shape and feeding traces independently support the same motion. They test physical plausibility rather than replaying an extinct animal's choices.

What remains unknown

Most dinosaurs have neither gut contents nor an attributable coprolite. Even exceptional preservation retains only part of a meal. Lips, tongue, salivary glands, microbiome and most of the intestine normally disappear. Biomechanics estimates possible loads but cannot observe choice.

Exact plant lists, the share of carrion, seasonal switches and differences between sexes usually remain unknown. CT scanning reveals replacement teeth, three-dimensional microscopy quantifies wear, and chemical analysis separates residues, but each new technique still needs a well-documented fossil context.

Chemical signals in enamel and bone can sometimes compare trophic levels or movement through landscapes, yet fossilisation may change the original composition. Even where a signal survives, it more often distinguishes broad ecological patterns than a named food item. Future advances will come from combining exceptional direct fossils with carefully sampled series rather than expecting one technique to produce a complete menu.