AncestryDNA
Largest matching pool in consumer testing.
Tree, ThruLines and record integration.
No health reports and no competitor uploads.
Largest matching pool in consumer testing.
Tree, ThruLines and record integration.
No health reports and no competitor uploads.
Deepest European and Jewish reference coverage.
Free uploads from other services.
Smaller US pool than Ancestry.
Chromosome browser and haplogroup labels.
Solid mixed-ancestry estimates.
Smaller pool, and an ownership change in 2025.
Unusually fine resolution inside Britain and Ireland.
Clear methodology write-up.
Small pool limits relative matching.
Real Y-DNA and mtDNA products for single lines.
Built for surname and lineage projects.
Small autosomal pool and a steep curve.
This is why the largest database keeps winning recommendations despite not being the most feature-rich product. Match quality is a numbers game at the close-relative level and a tooling game at the distant level, and you need both: a pool big enough to produce a fourth cousin, and a tree system capable of placing that cousin somewhere.
Smaller vendors compete on specificity. MyHeritage is stronger for continental European and Jewish ancestry because its reference panels are deeper there. Living DNA offers unusually fine resolution inside the British Isles, where larger vendors tend to collapse distinct regions into one label. FamilyTreeDNA sells Y-DNA and mitochondrial tests that trace a single line rather than estimating overall composition, which is a genuinely different product for a genuinely different question.
None of these are better kits. They are better fits for narrower questions.
If you already have raw data, the cheapest useful move is not another kit. MyHeritage accepts free uploads from other services and GEDmatch and similar analysis tools accept uploads as well. Uploading costs nothing, adds a second matching pool, and takes minutes. The only thing it does not give you is new sequencing, which is rarely what a second purchase was going to provide anyway.
We do not sequence kits or verify individual ethnicity estimates. We score what is checkable: whether the laboratory holds accreditation that can be looked up, whether marketing claims are specific enough to be falsifiable, how large and usable the matching pool is, and what the policy commits to on retention and deletion. A high score means better documentation and a stronger pool. It does not mean your percentages are correct.
The scores below compare these kits against the same evidence and governance topics. A high score indicates stronger documentation, a larger usable pool, or clearer data commitments. It does not mean an ethnicity estimate is accurate for your specific family.
This measures whether the product offers health reporting at all, and if it does, whether the scope is stated and authorized rather than implied. It matters because ancestry buyers frequently assume health insight is included, and most ancestry-first products deliberately exclude it.
This measures whether a customer who discovers something unexpected, such as a misattributed parentage event, has a route to a qualified human. It matters because surprise findings are common in ancestry testing and there is often no support pathway at all.
This measures whether consent for testing is separated from consent for research, and whether the default favors the customer. It matters because ancestry buyers are usually thinking about ethnicity, not about what they are agreeing to.
This measures whether the laboratory is named and holds accreditation that can be independently checked. It matters because the customer never sees the lab, only a dashboard.
This measures whether headline claims about regions, accuracy, and database size are specific enough to check. It matters because 'thousands of regions' is a marketing number, not an accuracy figure.
This measures the size of the matching pool and how usable a match becomes, including tree integration and record tools. It matters because this is the variable that actually determines whether the test produces a finding.
This measures whether the company explains its reference panels and states plainly that estimates are estimates. It matters because unexplained regional splits get treated as facts by customers who have no way to audit them.
This measures what the policy commits to on retention and whether deleting an account also destroys the physical sample. It matters because genetic data cannot be reissued the way a password can.
This measures ownership history and whether customers were told what happens to their data in a change of control. It matters because an ancestry account is a long-term relationship with a data custodian.
This measures whether research participation is opt-in and whether third-party sharing and law-enforcement requests are addressed specifically. It matters because these clauses are the ones customers most often assume rather than read.
For ancestry work specifically, AncestryDNA is the strongest product because it combines the largest pool with the deepest tooling on top of it. Ethnicity estimates run at a fine regional grain, and the integration with historical records is what separates a match from a research lead. The absences are deliberate: no health reporting, and no acceptance of uploads from other services, which makes it a starting point rather than a hub.
Anyone whose main goal is finding relatives or extending a family tree.
Buyers who want health information, or who already tested elsewhere and want one account to consolidate everything.
With no health claims to defend, the interpretive risk is low. Governance is conventional: documented deletion, a separate instruction needed to destroy the physical sample, and stable ownership.
MyHeritage is the best ancestry kit for anyone whose family is continental European, and the best second database for everyone else. Reference coverage for European and Jewish populations is deeper than the larger US vendors, the kit costs less at list, and free uploads mean a customer with raw data from elsewhere can join the pool without buying anything.
European family trees, budget buyers, and anyone extending an existing test result.
Someone whose only goal is a close US relative match.
Health add-ons sit outside the FDA-authorized category and should be treated as wellness content. Governance is unremarkable in the good sense, with documented deletion and no ownership disruption.
The ancestry-only tier at 23andMe is a competent product with two features its rivals lack: a chromosome browser and Y/mtDNA haplogroup labels. Estimates for mixed ancestry are strong. The pool is roughly half of Ancestry's and has been shrinking rather than growing, and the 2025 bankruptcy and asset sale are material context for anyone treating this as a long-term data relationship.
Buyers who want a chromosome browser and haplogroup detail, and who have read the current terms.
Buyers whose priority is the largest possible match pool, or who are uncomfortable with the ownership history.
Ancestry-only does not require engaging with health reporting. Governance is the weak point: an ownership change under bankruptcy court supervision and a documented breach belong in the decision.
Living DNA exists for one narrow question: what part of Britain or Ireland does my family come from? Where the large vendors collapse England, Scotland, Wales and Ireland into broad labels, Living DNA splits them further. For that specific job it is the better product. For anything else, the small database is a real limitation.
Buyers with British or Irish ancestry who want regional texture.
Anyone buying primarily to find relatives.
Health content is minimal and should not drive the purchase. Methodology transparency is stronger than the company's size would suggest, and deletion is straightforward.
FamilyTreeDNA's value is that it sells lineage tests rather than only an ethnicity estimate. Y-DNA traces the direct paternal line and mtDNA the direct maternal line, which is a different question from overall composition and one the big three largely do not answer. The name says it: this is a tool for people already doing genealogy.
Genealogists working a specific paternal or maternal line.
First-time buyers or anyone wanting health insight.
No health claims, so no interpretive risk. The policy is written for a genealogy audience and is denser than consumer-facing kits.