How Accurate Are Dog DNA Tests? Vendor Claims, Independent Studies and Health Results
Dog DNA tests are good at confirming a registered purebred and much less consistent with mixed-breed dogs, and no company publishes how its accuracy figure was measured. In a 2024 JAVMA study of 12 registered purebreds, five of six consumer tests always found the dog's registered breed, while the sixth appeared to rely on the photo. For mixed breeds, results depend on which breeds are in each company's reference panel, so two companies can report different mixes for the same dog. Health results are screening for gene variants, not a diagnosis: carrying a variant does not mean a dog will get the disease.
What the companies claim
Every company makes an accuracy claim, and none of them explains it. Here is what each says, checked on 10/09/2026:
| Company | What it says about accuracy | Method published? |
|---|---|---|
| Embark | "99% accuracy" for breed mix on its homepage and "99% breed accuracy" on its product pages (10/14/2026); no accuracy figure for health tests on the product, FAQ, genetic-health, or science pages we opened that day | No validation study found |
| Wisdom Panel | FAQ: "Because of the potential variation in DNA quality, we're unable to provide a definitive rate of accuracy at this time." Its blog says mixed-breed error rates were "2-4x lower than the industry standard". | Points to a white paper we could not open; a 2021 press release said ">98%" |
| DNA My Dog | "We test over 350 breeds with over 99% accuracy" | No |
| Orivet | No accuracy figure on the product pages read | No |
Two numbers circulate online that the sources don't support. A "98.7% accurate" figure attributed to Wisdom Panel comes from a 2022 Science study by Morrill et al., which measured a research pipeline, not Wisdom Panel. And Wisdom Panel's ">98%" is from a July 2021 press release; it is not on the current Breed Discovery page. For which kit to buy on panel size, price, and results time, see the best dog DNA tests.
Geno Pet Dog Breed Identification Test Kit
Dog DNA test · No subscription · first year $95What independent studies found
Three peer-reviewed studies are useful here, and each measured something different.
Rando et al., JAVMA 2024 (doi 10.2460/javma.23.07.0372) sent samples from 12 registered purebred dogs of 12 breeds to six companies: Embark, Wisdom Panel, Orivet, Darwin's Ark, DNA My Dog, and Accu-Metrics. Half the dogs were submitted with photos of a different dog. Five of the six tests always included the registered breed. Accu-Metrics matched none, and the authors wrote that "one test on the market is relying on the photograph more than the DNA sample." The authors are careful about scope: "this study was not designed to evaluate the accuracy of the tests' genetic ancestry predictions", and how vendors make their estimates "and their accuracy remains largely unknown". Mixed breeds were not tested.
Sexton et al., Scientific Reports 2025 (doi 10.1038/s41598-025-15095-5) compared owners' beliefs with genetic results for 5,673 dogs in the Dog Aging Project. "80% of dog owners identified their dog's breed as matching the genetic reports." When owners of single-breed dogs disagreed, the usual reason was that the dog's breed was "not included in the reference panel". The project's panel covers 114 breeds, far fewer than commercial tests, so this is a lesson about panels, not a grade for any company.
Morrill et al., Science 2022 (doi 10.1126/science.abk0639) ran a research pipeline on a 101-breed panel. The top breed it called matched the owner-reported breed "in 98.7% of dogs described as registered purebred" (304 dogs) and in 85.8% of dogs whose owners reported a single breed (885 dogs). In simulated mixes it "calls breeds comprising >5% ancestry accurately but misses lower-frequency breeds". It is a research tool, so its numbers do not transfer to any kit you can buy.
Why two companies can give different results for one mixed-breed dog
Each company compares your dog's DNA with its own reference panel of dogs of known breed, and the panels differ: Wisdom Panel says 430+ breeds, Embark 400+, and DNA My Dog 350+ (vendor figures, 10/09/2026). If an ancestor's breed isn't in a panel, the test has to assign that DNA to the closest breeds it does have. Small shares of ancestry are also the hardest to call, which is what the Morrill study found below 5%.
Journalists have tested this directly. In a 2023 CBC Marketplace report (03/04/2023), as summarized by Whole Dog Journal on 03/09/2023 (cbc.ca did not open to our tools), the same mixed-breed dog got different answers from different companies: a dog from Turkey came back "100% West Asian Village Dog" from Embark and a five-breed mix from Wisdom Panel. The purebred Great Dane in the same test was called "100% Great Dane" by both Embark and Wisdom Panel, while Accu-Metrics and DNA My Dog "failed to accurately identify the purebred Great Dane".
The same kind of split shows up in owner reports. A 2021 GoPetFriendly comparison of one dog found Wisdom Panel reported 13 breeds and Embark 7, with 16.6% labeled "Supermutt" by Embark, meaning DNA too mixed to assign. That report is more than two years old and both companies have since changed their panels. Recent one-star Trustpilot reviews of Embark include "They claim my shelter dog is 100% standard poodle." (04/13/2026). Single reviews prove nothing alone, but they match what the studies describe.
The human-swab tests
Reporters have also sent human cheek swabs to dog DNA companies to see whether a lab rejects DNA that isn't from a dog.
- CBC Marketplace, 2023: for a human sample, Embark and Wisdom Panel reported that "no dog DNA was present", while Accu-Metrics and DNA My Dog returned "various dog-breed mixes" (Whole Dog Journal's summary of the report, 03/09/2023).
- WBZ I-Team, CBS Boston, 03/13/2024: Orivet said the sample "failed to provide the data necessary to perform the breed ID analysis", and Wisdom Panel said it didn't provide "enough DNA to produce a reliable result". DNA My Dog reported the reporter as "40% Alaskan Malamute, 35% Shar-Pei, and 25% Labrador." Embark was not part of this test.
DNA My Dog told Global News on 03/15/2024 that the test had run on an "old testing platform" since replaced with a "much more robust and highly-sensitive system". A human-swab test only shows whether a lab screens out non-dog DNA; it says nothing about breed accuracy for real dogs.
Health results: screening, not diagnosis
Nothing on this page is veterinary advice; see our medical disclaimer. A positive health result is a reason to talk with your vet, who may order confirmatory tests.
The vendors and researchers agree on the limits:
- Wisdom Panel's own scientists, in a 2023 PLoS Genetics paper on 1,054,293 dogs (doi 10.1371/journal.pgen.1010651), wrote that "some potential for individual spurious false positive test results exists" and cautioned "against making conclusive health care or breeding related decisions" from the test alone for many variants in mixed breeds. All authors were Wisdom Panel employees.
- A gene variant is not the disease. The 2009 PNAS paper that linked an SOD1 variant to degenerative myelopathy (doi 10.1073/pnas.0812297106) found that "the penetrance among A/A homozygotes is incomplete" and that "several of the aged A/A homozygotes were symptom free". NC State's veterinary hospital puts it plainly: "not all genetically at risk dogs will develop the disease."
- The WSAVA Hereditary Disease Committee's position paper says a variant can sit in several breeds "but not necessarily be associated with disease in all of them. This means that a DNA test for one breed may not be appropriate for others." It asks that results be interpreted "to avoid improper decisions regarding breeding, treatment or euthanasia being made."
- A 2018 comment in Nature by Moses, Niemi, and Karlsson argued that "as few as 1 in 100 dogs that test positive for this common mutation will develop the very rare disease" and that "Neither their accuracy nor their ability to predict health outcomes has been validated" (as quoted on Harvard Medical School's Center for Bioethics page of 07/27/2018, read 10/14/2026; the Nature text beyond its opening paragraph is paywalled).
Drug-sensitivity markers are where a result is most directly useful. Washington State University, which developed the MDR1 test, says dogs with the variant "may have severe and life-threatening reactions" to some drugs and that "the only way to know if an individual dog has the mutant MDR1 gene is to have the dog tested." Share an MDR1 result with your vet before any treatment.
Who checks the labs
No one, in a binding sense. No federal agency reviews dog DNA tests before sale; Embark's help center notes the FDA does not regulate genetic tests for companion animals, and says its own tests run in a CLIA-accredited lab. Wisdom Panel, Orivet, and DNA My Dog do not publish lab accreditation on the pages I read. The International Partnership for Dogs runs a voluntary database, Harmonization of Genetic Testing for Dogs, where labs list their own tests and the partnership rates how relevant each test is to each breed, more than 2,000 test-and-breed combinations (IPFD's figure as of January 2022, read 10/09/2026). The labs supply the data themselves, and the partnership does not certify them. Embark's Adam Boyko told NPR in 2018: "We don't want it to be the Wild West."
How to read your own result
- A purebred confirmation from a major company is the most reliable kind of result, per the JAVMA study.
- Breed shares under about 5% are the least reliable part of a mixed-breed report.
- A "village dog" or "Supermutt" share means the DNA did not match a breed in the panel. It is not an error.
- A health "at risk" finding goes to your vet, not into a decision on its own. Ask whether the variant is known to cause disease in your dog's breeds.
- Results from two companies will not match exactly for most mixed breeds, because their panels differ.
Panel size, health tests, price, and results time for each kit are compared in the best dog DNA tests.
Questions people ask
Which dog DNA test is the most accurate?
There is no independent ranking. In the only peer-reviewed head-to-head we found (JAVMA, 2024), five of six tests, including Embark, Wisdom Panel, Orivet, and DNA My Dog, always found the registered breed of 12 purebred dogs; Accu-Metrics did not. Mixed breeds were not tested. Panel size and price compared: best dog DNA tests.
Is Embark accurate?
Embark claims "99% accuracy" for breed mix but publishes no validation study behind that number. In the 2024 JAVMA study, Embark always included the registered breed for 12 purebred dogs. For mixed breeds, owners on Trustpilot sometimes dispute results, as they do for every company.
Is Wisdom Panel accurate?
Wisdom Panel's FAQ says it is "unable to provide a definitive rate of accuracy at this time", though it claims mixed-breed error rates "2-4x lower than the industry standard". Like Embark, it found the registered breed for all 12 purebred dogs in the 2024 JAVMA study. More in the Wisdom Panel review.
Can a dog DNA test be fake?
One test appeared to be. In the 2024 JAVMA study, Accu-Metrics matched none of the registered breeds, and the authors concluded it was "relying on the photograph more than the DNA sample." Separately, DNA My Dog returned dog breeds for human swabs in reports by CBC (2023) and CBS Boston (2024).
Can a DNA test tell if my dog will get sick?
No. It can tell you whether your dog carries a variant linked to a condition. Many variants have incomplete penetrance, so some dogs carrying them never show signs; Wisdom Panel's own researchers warn of possible false positives. Take any "at risk" result to your vet.