How Dog DNA Could Help Solve Crimes
A UK dog DNA reference database could help investigators link a dog to crime-scene evidence—or rule one out. Here’s how the forensic science works and where it stops.
Key takeaways
- Forensic scientists at King’s College London and the Royal Veterinary College have built a UK reference database of dog DNA profiles from more than 1,200 dogs across 80 breeds.
- It helps experts judge how meaningful a match is between DNA left at a scene and a known dog, and it can just as easily help rule a dog out.
- Separate data for the UK’s most common pedigree breeds makes those statistics more accurate when a dog’s breed is known.
- It is a UK research and casework tool. It is not a register of every dog, not a consumer ancestry test, and not proof of a crime on its own.
When police investigate an attack, a theft or an assault, the people involved get most of the forensic attention. The dogs, which shed hair and leave saliva behind, usually don't. A team of UK forensic scientists wants to change that.
Researchers at King's College London (KCL) and the Royal Veterinary College (RVC) have assembled a reference database of canine DNA drawn from the UK's dog population, covering familiar breeds such as Labradors, Cocker Spaniels, Dachshunds and Pugs. The Guardian reported on the work on September 25, 2026, and KCL published its own announcement on September 28. The underlying peer-reviewed study appears in Forensic Science International: Genetics. It was first published online in September 2025 and sits in the journal's February 2026 volume.
"Dogs are such a big part of our lives – but often overlooked in forensics," Prof. Denise Syndercombe Court, the KCL forensic scientist who leads the work, said in the university's announcement.
What's actually in the database?
The team genotyped more than 1,200 dogs from 80 breeds. Most samples were blood taken during veterinary checks at the RVC, which says it has built its canine DNA archive over 25 years. Public donations and a contribution from the University of Edinburgh filled out the set.
Researchers created breed-specific allele-frequency databases for several common breeds within a broader UK dataset covering 80 breeds and more than 1,200 dogs. Those breed-specific frequencies are used when the breed is known; otherwise, analysts use the whole-population database with a conservative correction.
Each dog was profiled at 13 short tandem repeat (STR) markers. These are stretches of DNA where a short sequence repeats a variable number of times, so different dogs tend to carry different counts. The specific marker set, called the CaDNAP panel, was validated for forensic use by the Canine DNA Profiling Group, an international group founded in 2003. It is the dog equivalent of the approach long used in human DNA profiling.
One point is easy to miss: the database is a record of how common each DNA variant is among UK dogs. It is not a lookup list of named pets.
How does dog DNA become evidence?
- Recover it. Saliva on a bite wound or torn clothing, blood or hair at a scene may carry canine DNA. KCL notes that dogs can be directly involved, by attacking people or livestock, or can be bystanders whose traces help link a human suspect to a place.
- Compare it. Scientists profile the trace and compare it with a reference sample from a specific dog, such as a mouth swab.
- Weigh the match. If the profiles match, the key question is how likely that result would be if the trace had come from some other, unrelated dog. The answer is expressed as a likelihood ratio, and the database supplies the variant frequencies needed to calculate it.
- Or rule it out. If the profiles differ, the suspected dog can be excluded, which can matter just as much to an owner whose dog has been wrongly blamed.
Why breed makes the math tricky
Pedigree dogs are, by design, closely related within their breed. The authors note that a UK-wide database "combines multiple different inbred breeds," so two unrelated dogs of the same breed share more DNA than two unrelated people typically would. Left uncorrected, that could make a coincidental match look rarer than it really is.
To compensate, the researchers recommend a conservative statistical correction (a theta, or FST, value of at least 0.2) when using the whole-UK data, and 0.05 when a breed-specific database applies. In test calculations for Pugs and French Bulldogs, a full 13-marker match still produced likelihood ratios in the billions even after the stronger correction. The authors also caution that Pugs are so inbred that even a 0.2 correction may not be conservative enough for that breed.
Their recommendation is practical: if the dog's breed is known and it is one of the common breeds covered, use the breed-specific data. If it isn't known, use the whole-population data with the larger correction.
Can DNA reveal what breed a dog was?
Sometimes, with caveats. KCL says scientists can use the data "to infer the likely breed or dog type." The team's project page says more than 60 of the 80 breeds studied carry a characteristic set of markers that could provide "useful intelligence" in an investigation.
But the new peer-reviewed paper reports variant frequencies and match statistics; it does not itself test breed assignment, and mixed-breed dogs complicate any breed estimate. A likely breed is a lead for investigators. It is not the same as identifying an individual dog.
How it has been used so far
In KCL's announcement, Syndercombe Court described two cases. In one, a Staffordshire Bull Terrier was suspected of attacking a police officer, and the owner denied it. A mouth swab from the dog produced a result about one billion times more likely if the crime-scene stain came from that terrier than from another dog.
In the other, a man accused of injuring someone said the other man's dogs had been set on him. Damage to his T-shirt carried likely canine saliva with DNA from two dogs, and that evidence supported his defence. It is a useful reminder that forensic evidence can help clear people, not only implicate them.
KCL says the team has also identified dogs that attacked livestock, and that two UK police forces have worked with the researchers to bring dog DNA analysis into their investigations. The forces were not named.
Shop the Look
As an Amazon Associate I earn from qualifying purchases.
Not everyone expects fast adoption. Nicholas Dawnay, a forensic scientist at Liverpool John Moores University who was not involved, told The Guardian that police budgets are stretched and that without effective funding the benefits "may struggle to be realised."
What this doesn't mean
- It isn't a register of every dog. The database was built from roughly 1,300 veterinary and donated samples. It does not require owners to register anything.
- It can't identify any dog from any trace. Results depend on recovering usable DNA and on having a known dog to compare against. Syndercombe Court said the team has the tools to analyse dog hairs left at crime scenes but has not yet done so in a case.
- A match isn't proof of a crime. DNA can indicate where biological material likely came from. It cannot say by itself who did what, or when, and it has to be weighed alongside how evidence was collected and handled and everything else in a case.
- It isn't a consumer dog DNA test. At-home kits are sold to tell owners about ancestry. This is a forensic statistics reference built for casework, and it answers a different question.
- It is UK-specific. The frequencies describe dogs in the UK. The reviewed sources do not establish a comparable national dog DNA database in the United States.
What comes next
Syndercombe Court told KCL she wants to expand the database to include dangerous dog breeds, to help in cases where people have been attacked or killed by dogs. The team's project page also describes parallel work on dogs' mitochondrial DNA.
For dog owners, the takeaway is less about crime than fairness. When a dog is accused of something, careful DNA work can confirm what happened, and it can also show that a dog wasn't there at all.
Sources
- Giangasparo F, Ballard D, Colligan R, Anderson F, Chung C-Y, Catchpole B, Syndercombe Court D. Characterisation of the UK domestic dog population for identification purposes using a forensically validated set of 13 autosomal STR markers. Forensic Science International: Genetics 81:103364. Published online September 19, 2025; volume dated February 2026.
- King’s College London. Cracking crimes with canines: forensics experts build database of dog DNA. September 28, 2026.
- King’s College London. Canine individual and breed identification research project page.
- Davis N. DNA database of UK dogs could help police collar criminals. The Guardian, September 25, 2026.
About the author
Stephen Gill
Stephen Gill is the founder of Inspired by Tyson, sharing thoughtful guidance and stories for people who want the best for their dogs.
