Most dogs in Britain are not purebred, and most owners of a crossbreed are working from a guess someone made at a rehoming centre. The tool above narrows the field from physical traits. This guide explains which traits are actually informative, how often appearance-based identification is wrong, what a DNA test will and will not tell you, and the one situation in which getting it wrong has legal consequences.

What breed is my dog?

Breed identification from appearance works by elimination. No single feature identifies a dog, but a combination of six or seven — adult size, head shape, ear carriage, coat type, tail set and body proportion — narrows several hundred possibilities to a handful.

The catch is that crossbreeding does not blend traits the way people expect. A first cross between a Labrador and a Poodle does not produce something halfway between them; it produces a range across the litter, with some pups taking the coat of one parent and the head of the other. By the second or third generation, the visible relationship to either founder breed can be very weak while the DNA is unchanged.

The traits that carry the most information

1. Adult size and build

The most powerful single filter, because it immediately removes most of the field. Record height at the withers — the ridge between the shoulder blades — as well as weight, since two dogs of the same weight can have very different frames. Note whether the build is square (height roughly equal to body length, as in a Boxer), long-bodied (Dachshund, Corgi, Basset) or leggy (sighthounds).

2. Head and muzzle shape

This is the trait that survives crossbreeding best, and the one most useful for identifying an unknown mix.

• Brachycephalic — flat, shortened muzzle: Pug, French Bulldog, Bulldog, Boxer, Shih Tzu, Pekingese, Boston Terrier.
• Mesocephalic — balanced, medium muzzle: Labrador, Beagle, Spaniel, most terriers, Staffordshire Bull Terrier.
• Dolichocephalic — long, narrow muzzle: Greyhound, Whippet, Collie, Borzoi, Afghan Hound, German Shepherd.

Also note skull width between the ears, the stop (the step between muzzle and forehead), and jaw shape — an undershot jaw points strongly at bull breeds or brachycephalic ancestry.

3. Ear carriage

Ear set is highly heritable and surprisingly informative. Prick ears standing fully erect suggest spitz types, German Shepherd, Corgi or terrier ancestry. Drop ears hanging flat point to gundogs and hounds. Rose ears, folded back and slightly outward, are characteristic of sighthounds and bull breeds. Button or semi-erect ears — standing then folding forward — are common across terriers and collies.

4. Coat type

Coat is under relatively simple genetic control, which is why DNA tests predict it well and why it is unreliable as a visual clue on its own. Short smooth coats, double coats with a dense undercoat, wire coats, silky coats and curly or wool coats each point to different groups — but a curly coat in a mix usually means poodle somewhere, and little else.

5. Tail

Note set (high or low on the croup), carriage (over the back, level, or tucked) and shape (straight, sabre, sickle, curled, screw, or naturally bobbed). A tightly curled tail over the back suggests spitz breeds; a low-set tail carried level suggests hounds and gundogs; a screw tail is a bulldog marker.

6. Colour and markings

The least reliable of the six. Merle, brindle, saddle-back tan points, ticking and piebald are each associated with particular groups, but colour genetics are simple enough that they cross freely. A black-and-tan dog is not necessarily a Rottweiler cross.

7. Behaviour and movement

Often more informative than any single physical feature. Herding behaviour — circling, eye-stalking, heel-nipping — is highly specific. So is a retrieve-and-carry habit, a hound's nose-down tracking, a terrier's digging and prey-shaking, and a sighthound's explosive sprint-then-sleep pattern.

Narrowing it down by breed group

Common UK crossbreeds and what they look like

Britain's crossbreed population is dominated by a small number of combinations, which is useful — the prior probability that an unknown curly-coated medium dog is some sort of doodle is high.

Why visual identification gets it wrong

Studies comparing shelter staff's visual breed assignments with DNA results have repeatedly found poor agreement. In several, the breed named on the kennel card was not among the breeds the DNA test detected in the majority of dogs. Experienced staff disagreed with each other as often as they agreed.

There are structural reasons for this.

• Appearance is controlled by a small number of genes. Coat, colour, ear set and muzzle length are governed by relatively few loci, so a dog can look strongly like one breed while carrying little of its ancestry.
• Mixed-breed dogs regress towards a generic type. After a few generations of random mixing, dogs converge on a medium-sized, short-coated, prick-eared, tan or brindle type — the “village dog” phenotype seen worldwide.
• Humans anchor on one feature. A brindle coat leads to “Staffie cross”; a black-and-tan saddle leads to “German Shepherd cross”; and once the label is applied, other features get read to fit it.
• Labels carry consequences. A breed label affects adoption speed, insurance quotes, and in some cases housing — which creates pressure on the guess in both directions.

Dog DNA tests in the UK

A dog DNA test uses a cheek swab and compares hundreds of thousands of genetic markers against a reference panel of known purebred dogs. The main kits available in Britain are Wisdom Panel and Embark, with several smaller providers.

What they do well. Identify the major breeds in a mix, predict adult size in a growing puppy with reasonable accuracy, and — the genuinely valuable part — flag single-gene health risks. The MDR1 mutation, common in collies and their crosses, causes severe and occasionally fatal reactions to several routine veterinary drugs including some wormers; knowing about it before an anaesthetic is worth the price of the kit on its own.

What they do less well. Small percentages are unreliable — a “3 per cent Chow Chow” result usually means the algorithm could not place a fragment, not that a Chow was involved. Results differ between companies for the same dog because the reference panels differ. Village-dog and street-dog ancestry, common in imported rescues from Romania, Cyprus and Greece, is often reported as a scatter of unrelated breeds. And breed percentages predict behaviour only weakly: large studies have found that breed explains only a small fraction of behavioural variation between individual dogs.

What knowing the breed actually changes

Size is the one that touches the budget hardest. If you have a growing puppy of unknown parentage, the predicted adult weight from a DNA test or a confident visual identification will change your food, insurance and grooming estimates substantially — the puppy cost calculator shows by how much.

Breed type and UK law

This is the one area where getting identification wrong has serious consequences. Under the Dangerous Dogs Act 1991, certain types of dog are prohibited in the UK — the Pit Bull Terrier type, Japanese Tosa, Dogo Argentino and Fila Brasileiro — and the American Bully XL was added to that list, with the ban taking full effect in England and Wales from February 2024 and in Scotland later that year. Northern Ireland has its own timetable and guidance.

Three points owners consistently misunderstand:

• It is type, not breed. A dog is assessed against a written conformation standard — measurements, proportions, head and body characteristics. A dog with no pit bull or bully ancestry at all can be judged to be of type, and a dog with such ancestry can fall outside it.
• A DNA test is not a defence. Genetics do not determine the assessment; physical conformation does.
• Exempted dogs live under conditions. Existing XL Bullies could be kept under a Certificate of Exemption, which requires neutering, microchipping, muzzling and leads in public and secure containment at home, with the keeper aged over 16. Those conditions have continued to be adjusted — from 1 July 2026 the third-party public liability insurance requirement for a Certificate of Exemption is being removed, and from 1 November 2026 it becomes an offence to leave a child of 12 or under alone with an XL Bully without adult supervision.

If you have any reason to think your dog might be assessed as a banned type, take advice before acting. The RSPCA, Blue Cross, Battersea and the Royal Kennel Club all publish current guidance, and gov.uk and nidirect carry the official rules for each nation. Do not rely on a breed identification tool, including this one, for a legal question.

Reading a rescue dog's breed label

Rehoming centre labels are informed guesses made quickly, often by someone who has seen thousands of dogs but has no history for this one. Treat them as a description of appearance, not ancestry.

More useful than the label:

• The behavioural notes. “Chases bikes”, “carries a toy everywhere”, “follows a scent and goes deaf” tell you more about what you will live with than any breed name.
• Adult weight and height, if the dog is grown. This drives cost and practicalities.
• Coat type, because it decides grooming spend — see the grooming cost estimator.
• Country of origin. Imported street dogs need specific checks: UK vets screen for leishmaniasis, ehrlichiosis, babesiosis and heartworm in dogs from southern and eastern Europe.

When breed does not matter at all

Research on tens of thousands of dogs has found that breed accounts for only a modest share of the variation in individual behaviour. Breed shifts the odds — a Border Collie is more likely to be biddable and noise-sensitive — but it does not determine the dog in front of you. Early experience, training, health and temperament vary enormously within every breed.

So use breed to inform health screening, adult size, coat care and exercise planning. Do not use it to predict whether your particular dog will be good with children, easy to train or comfortable alone. Those come from the individual, and from what you do with them.

Frequently asked questions