AncestryDNA
Largest consumer matching pool by a wide margin.
Tree, record and ThruLines tools turn matches into research.
No health reports, and no uploads from other services.
Largest consumer matching pool by a wide margin.
Tree, record and ThruLines tools turn matches into research.
No health reports, and no uploads from other services.
Cheapest of the major three at list price.
Strong reference coverage for European populations.
The only major kit that accepts free uploads from rivals.
The only mainstream kit with FDA-authorized health reports.
Chromosome browser and haplogroup detail included.
Bankruptcy and a 2023 breach weigh on the governance score.
Y-DNA and mtDNA testing for deep paternal and maternal lines.
Best tooling for serious genealogists.
Smallest pool and the steepest learning curve here.
Fine-grained regional breakdowns, especially within Britain and Ireland.
Lower price point than the US majors.
Smaller database limits relative matching.
Whole-genome sequencing rather than a genotyping array.
App marketplace for third-party analysis.
The most data for the most work, and the least hand-holding.
That is why two kits at very different prices can return similar-looking percentages and still be very different purchases. The percentages are an estimate produced by comparing your markers against a reference panel; the panel and the algorithm behind it vary by company, which is why the same person can appear 62% one thing at one service and 58% at another without either being fraudulent.
AncestryDNA's advantage is scale. A pool in the tens of millions means more chances of a usable match, and the integration with historical records turns a match into a name and a place. What it does not sell is health reporting, and it does not accept raw uploads from other services.
23andMe's advantage is regulatory. Its health reports carry FDA authorization, which no other mainstream consumer kit can claim. The offset is corporate: a 2025 bankruptcy, an asset sale, and a 2023 breach affecting millions of accounts. That does not make the reports invalid, but it does change the calculus for anyone whose main concern is who holds the data in five years.
MyHeritage's advantage is price and flexibility. It is the cheapest of the three at list, it is strong on European reference populations, and it accepts free uploads, which makes it the natural second database rather than a replacement for a first one.
We do not run our own sequencing, and we do not tell you what a result means for your health. The scores below measure verifiable properties: laboratory accreditation, whether claims are backed by an authorization or a certification that can be looked up, how clearly consent and research participation are separated, and what the retention and deletion policy actually commits to. A high score means better documentation and governance. It does not mean the kit is more accurate, and it does not mean a health result is something to act on without a clinician.
The scores below compare these kits against the same health, evidence, and governance topics. A high score means stronger documentation or verification within that topic. It does not mean the product is risk-free, and it does not mean a health report should be read without professional context.
This measures whether a health report states which conditions are covered, whether that coverage rests on an authorization that can be verified, and whether the product is explicit that a result is risk information rather than a diagnosis. It matters because 'health DNA test' is used loosely in marketing while the regulated scope is narrow and specific.
This measures whether a customer who receives a concerning result has a route to a qualified human, whether that route is included or sold separately, and whether the company tells you to seek one. It matters because a raw probability without interpretation is where consumer genomics most often goes wrong.
This measures whether consent for testing and consent for research are separate choices, whether the default is opt-in or opt-out, and how clearly the sample-destruction option is presented. It matters because bundled consent is the quiet mechanism by which genetic data enters research programs.
This measures whether the laboratory behind the kit holds accreditation that can be independently checked, and whether the company names it. It matters because the consumer sees a dashboard, not a lab, and accreditation is the main external signal that the pipeline is what it claims to be.
This measures whether headline claims map to something a customer can look up: an authorization, a certification, a published methodology. It matters because consumer genomics marketing routinely implies clinical certainty that no authorization supports.
This measures the size of the matching pool and how usable a match is once found, including tree tools and record integration. It matters because relative matching is a network problem, and a better algorithm on a small pool loses to a plain algorithm on a large one.
This measures whether the company explains its reference panels, states that estimates are estimates, and publishes how regions are defined. It matters because the same DNA can produce different regional percentages at different services without any of them being wrong.
This measures what the policy commits to on retention, whether account deletion also destroys the physical sample, and whether backups are addressed. It matters because genetic data is uniquely persistent: a leaked password can be changed, a sequenced genome cannot.
This measures ownership history, financial distress, and whether customers were told what happens to their data in a change of control. It matters because the 2025 bankruptcy showed that a data custodian can change without the customer doing anything at all.
This measures whether research participation is opt-in, whether third-party sharing is described specifically rather than gestured at, and whether law-enforcement requests are addressed in the policy. It matters because these are the clauses customers most often assume rather than read.
AncestryDNA wins on the one thing that compounds: other people. A matching pool in the tens of millions means more chances that a match is close enough to be useful, and the surrounding genealogy tools are what let a customer do something with a fourth cousin once the match appears. It sells no health reports, and it does not accept raw uploads from competitors, which makes it a first purchase rather than a hub.
Anyone whose primary goal is finding living relatives or breaking through a genealogical wall.
Anyone buying specifically for health risk information, or anyone who already tested elsewhere and wants to consolidate into one account.
The absence of health reporting is a limitation, not a failure: with no clinical claims to defend, the product carries less interpretive risk. On governance, the company's ownership has been stable and its deletion process is documented, though deletion is a request rather than a one-click control and the physical sample needs a separate instruction.
MyHeritage is the pragmatic second database. It costs least of the big three, its reference panels are strong for European populations in particular, and it accepts free raw-data uploads, which means a customer who already tested with Ancestry can extend their match pool without paying for a second kit. The matching pool is materially smaller than Ancestry's, so it works better as an addition than as a starting point for someone chasing a close relative.
European family trees, budget-first buyers, and anyone who already tested elsewhere and wants a free second pool.
Someone whose only goal is finding a close US relative, where pool size matters more than price.
Health add-ons exist but sit outside the FDA-authorized category, so treat them as wellness content rather than clinical reporting. Governance is unremarkable in a good way: documented deletion, no ownership disruption, and a policy that addresses research participation separately from testing consent.
23andMe is the only mainstream consumer kit whose health reports carry FDA authorization, and that distinction is real rather than semantic: it means named reports were reviewed against a regulatory standard instead of self-certified. The ancestry side is competent, with a chromosome browser and haplogroup detail its rivals lack. The problem is everything around the test. The company filed Chapter 11 in March 2025 and its assets were sold under court supervision, and a 2023 breach exposed data tied to roughly 6.9 million accounts. That history belongs in the buying decision, not in a footnote.
Buyers whose specific goal is FDA-authorized health and trait reporting, and who have read the current privacy terms rather than the pre-2025 version.
Anyone buying primarily to find relatives, or anyone who is not comfortable with a data custodian that has already been through a change of control.
The health reporting is the strongest in the category and the only part resting on an authorization. The governance picture is the weakest of the three: ownership changed under bankruptcy, the breach is documented, and research participation has historically been presented in a way that rewarded reading carefully. If you buy, download your raw data immediately and review whether you want the account to persist.
FamilyTreeDNA does something the big three largely do not: dedicated Y-DNA and mitochondrial tests for tracing a single paternal or maternal line rather than estimating overall ancestry. For surname projects and deep lineage work that is genuinely useful, and the tools assume a user who knows what a haplogroup is. The trade-off is a much smaller autosomal pool, so it is a specialist instrument rather than a first kit for most people.
Genealogists working a specific paternal or maternal line, or running a surname project.
First-time buyers, or anyone hoping for health insight.
No health claims, so no interpretive risk. Governance is conventional and the company has been stable, though the policy is written for a genealogy audience and is denser than the consumer-facing kits.
Living DNA's pitch is granularity inside the British Isles, where the big three tend to collapse several distinct regions into one label. For someone whose family is English, Scottish, Welsh or Irish and who wants more texture than 'Britain and Ireland', that specificity is the product. Relative matching is a secondary feature here, and the database is small enough that it should not be the reason to buy.
Buyers with British or Irish ancestry who care about regional detail more than match volume.
Anyone buying to find relatives, or anyone whose ancestry is primarily outside the British Isles.
Health content is limited and should not be the reason to choose this kit. The methodology documentation is better than its size would suggest, and the deletion policy is straightforward.
Sequencing.com sells whole-genome sequencing instead of the genotyping array the consumer kits use, then routes the resulting data through a marketplace of third-party apps. That is more raw information than any of the other kits provide, and it is also more rope: the apps vary in quality and in what they claim, and the company that sequenced your sample is not the company interpreting it. It became the practical whole-genome option after Nebula Genomics shut down in early 2025.
Users who want the underlying data and are comfortable evaluating analyses themselves.
Anyone who wants a finished, interpreted answer, or who is buying mainly to find relatives.
The health angle depends entirely on which third-party app you run, and those apps are not uniformly regulated, which is the central risk. On governance, portable data is a genuine advantage: you hold the file, so a change of vendor does not strand your results.