23andMe Review (2026)

23andMe is still the strongest mainstream DNA kit if you want ancestry and health screening in one account. The $99 Ancestry Service tier covers ancestry only; the $199 Premium tier is where the case for buying 23andMe over an ancestry-first kit actually starts, because that is where the FDA-authorized carrier and pharmacogenetics reports, plus selected FDA-reviewed health-risk reports, unlock. Skip 23andMe if your goal is family-history research. AncestryDNA is the better starting point for that.

This review covers the kit, the FDA pathway behind each report category, what the SNP-chip can and cannot tell you, the post-bankruptcy TTAM-era privacy picture, current pricing, and the cases where you should buy something else.

Key Takeaways

• 23andMe is worth buying for health screening plus ancestry. It is not the best first choice for genealogy-focused work.
• The entry price is $99. Premium is $199 and unlocks the FDA-authorized carrier and pharmacogenetics reports, selected FDA-reviewed genetic health-risk reports (including the BRCA1/BRCA2 Selected Variants report expanded through 510(k) clearance), and 23andMe Research–powered reports like the Type 2 Diabetes polygenic risk score that have not been reviewed by FDA. Total Health is $499.
• “FDA-reviewed” is category-by-category, not blanket. Carrier-screening and pharmacogenetics reports are FDA-authorized through De Novo decisions. BRCA selected variants and MUTYH were cleared through 510(k). Some predisposition reports, including the Type 2 Diabetes polygenic score, are 23andMe Research–powered and have not been reviewed by FDA. Ancestry, traits, and DNA Relatives sit outside FDA jurisdiction entirely.
• The BRCA report tests selected variants only. It screens 44 selected variants out of more than 1,000 known pathogenic BRCA1/BRCA2 variants; a negative result does not rule out hereditary breast or ovarian cancer risk.
• The 2025 ownership change matters. TTAM Research Institute completed the asset acquisition on July 14, 2025. The consumer service continues to operate.

Verdict and Who Should Buy It

23andMe makes the most sense when you want one account that combines ancestry with FDA-authorized carrier status and pharmacogenetics plus selected FDA-reviewed health-risk screening. No other mainstream consumer kit packages those categories alongside a polished ancestry product.

Buy Premium if you are curious about general health screening, planning a family and want carrier-status results before a clinician conversation, or interested in named pharmacogenes that affect common medications. The reports raise the right questions early; they do not answer them.

A different rule applies if you already have real reason to worry about hereditary cancer risk. If a first-degree relative or your own history puts BRCA- or MUTYH-related cancer concerns on your radar, start with a clinician or genetic counselor and ask about clinical-grade testing. A 23andMe result can inform that conversation, but the selected-variants panel is too narrow to use as reassurance. For APOE and late-onset Alzheimer’s, treat the result as a risk modifier to discuss with a clinician, not as a forecast.

Skip 23andMe if your priority is building a family tree, finding more relatives, or working with historical records; that job belongs to AncestryDNA. Also skip it if you expect a consumer DNA test to act like a diagnosis; the FDA, CDC, and 23andMe itself are explicit that results are screening, not medical decisions.

What You Get in the Kit

23andMe sends a saliva collection tube to your door. You spit, register the barcode, mail it back in the prepaid envelope, and results land in your account dashboard a few weeks later. The same sample is run through a CLIA-certified, CAP-accredited lab and feeds every report category your tier unlocks.

Ancestry Reports

Ancestry composition spans 4,500+ regions and links to DNA Relatives, the Family Tree builder, paternal and maternal haplogroups, Historical Matches, and DNA Relatives Clustering. The geographic detail is strong for casual and moderate exploration. It does not feed into a records-based workflow the way AncestryDNA does, but for buyers who came for health-plus-ancestry, the ancestry side holds up.

Ancestry, trait, and relative-finding features sit outside FDA jurisdiction. They are not FDA-reviewed, and we do not describe them that way.

Health Predisposition and Carrier Reports

Premium unlocks two FDA-authorized report families, carrier-screening (De Novo DEN140044, 2015) and Genetic Health Risk (De Novo DEN160026, 2017), plus a set of research-powered reports that have not been reviewed by FDA. The split matters, and 23andMe’s own healthcare-professional page draws the line in the same place. Both FDA-reviewed families have been expanded with later filings and added indications, which is why the current compare page frames health coverage as “45+ conditions” for Premium and “90+ conditions” for Total Health rather than a single “FDA-authorized report” headline count.

On the research-powered side, the dashboard also includes reports built by 23andMe Research that have not been reviewed by FDA. The Type 2 Diabetes polygenic risk score is the clearest example: a statistical risk estimate 23andMe developed in-house. It belongs in your decision math, but it does not carry the same regulatory bar as the carrier, GHR, BRCA, or MUTYH reports.

Pharmacogenetics

The pharmacogenetics reports were authorized under a 2018 De Novo (DEN180028, corrected 2019). Per the current 23andMe reports list, the Premium pharmacogenetics report covers three genes: CYP2C19 (3 variants, relevant to clopidogrel and certain PPIs), DPYD (2 variants, fluoropyrimidine chemotherapy toxicity), and SLCO1B1 (1 variant, statin-related side effects). The reports tell you which versions of those genes you carry. They do not tell you which medication is right for you, they do not test every relevant variant, and you should not start, stop, or change a medication based on them without a clinician.

Trait Reports

Trait reports are the social-sharing tier: cilantro aversion, ear-wax type, freckles, fear of public speaking, light-versus-deep sleep, and roughly 30 others. They are entertainment-grade and outside FDA review. Treat them as probabilities, not predictions.

First-party walkthrough planned: we will embed annotated dashboard screenshots of the ancestry composition wheel, health predispositions card, carrier-status card, and pharmacogenetics report in a subsequent update.

How Accurate Is 23andMe?

Accuracy depends on what you are asking the kit to do. The same SNP-chip that delivers a reliable carrier-status result for cystic fibrosis can completely miss a rare pathogenic BRCA variant a relative carries. Splitting the question by report type matters more than any single accuracy number.

How the SNP-Chip Works

23andMe uses a SNP genotyping array, a chip that reads roughly 600,000 pre-selected positions in your genome where humans commonly vary. That is fundamentally different from whole-genome sequencing, which reads every position.

For most consumer use cases (ancestry, carrier status for common conditions, named pharmacogenetic variants), the SNP-chip is the right tool. For ruling in or out a specific rare familial variant, it is the wrong tool. The BMJ analysis by Weedon and colleagues (2021) showed how often rare pathogenic variants go unseen or generate false positives on SNP-chip platforms, which is the single most important context for any “is 23andMe accurate” question.

Lab and Call-Rate Standards

23andMe runs samples through CLIA-certified and CAP-accredited laboratories and publishes a per-sample call-rate threshold above 99%. That call-rate is about the chip’s reliability at sites it is designed to read, not about how much of your genome is covered.

Ancestry Composition Accuracy

Ancestry estimates are strongest at the continental level (European, East Asian, Sub-Saharan African) and get fuzzier at country-level resolution, especially where the reference panel is thinner. Two siblings can get visibly different breakdowns because each inherits a different random half of each parent’s DNA. These are population-statistical estimates, not biological identity statements.

DNA Relatives Accuracy

DNA Relatives uses shared centimorgans (cM) of identical-by-descent DNA to predict relationships. Parent and child share roughly 3,400 cM. Full siblings share around 2,600 cM. First cousins land near 850 cM. By the fourth-cousin range, shared cM falls into a 20–60 cM window and many genuine relatives drop off the chart entirely. Close-family predictions are reliable; distant predictions are probabilistic.

Health and Trait Accuracy

For FDA-reviewed variants the company tests, internal lab accuracy is published above 99%. The catch is variant coverage. The BRCA selected-variants report is the cleanest example: the chip looks at a defined list and reports on those positions, and pathogenic BRCA variants outside that list are invisible to the test. The same principle applies to MUTYH and any selected-variants health-risk report. The research-powered polygenic risk reports, Type 2 Diabetes among them, are a separate accuracy question entirely, combining many small-effect-size variants into a statistical estimate without FDA review. Trait reports rely on the same kind of small-effect-size associations and should be read as tendencies, not predictions.

Health Reports in Detail

The Premium and Total Health dashboards stack four report families on top of the ancestry product. Each has its own FDA pathway (or, in the case of the polygenic risk reports, no FDA pathway at all) and its own ceiling. The CDC, MedlinePlus Genetics, and FDA are consistent on one point: these are screening tools that flag risk, not clinical diagnostics.

Carrier Status

Carrier-status reports trace back to the 2015 De Novo (DEN140044) and have been expanded by later filings. The current report set covers more than 40 conditions; medically high-impact entries include cystic fibrosis, sickle cell anemia, Tay-Sachs disease, Gaucher disease (Type 1), MCAD deficiency, Canavan disease, familial Mediterranean fever, Bloom syndrome, and Fanconi anemia (Type C). Most are autosomal-recessive: a positive result means you carry one copy, and the relevant risk case is a child with two carrier parents. Carrier reports work best as a starting point before discussing family planning with a genetic counselor; a positive result is not a diagnosis.

One condition readers often expect on this list lives elsewhere: HFE-related hereditary hemochromatosis is reported as a Genetic Health Risk under DEN160026 (HFE C282Y and H63D variants), not as a carrier-status result. The framing matters: hemochromatosis affects the person who carries the variants, not a future child of two carriers.

Health Predisposition Reports

The Genetic Health Risk family was authorized in 2017 (DEN160026) and now spans cardiovascular, neurological, and metabolic predispositions. Named reports inside the FDA-reviewed GHR bucket include:

• Late-onset Alzheimer’s disease (APOE ε4, a risk-modifier, not a guarantee)
• Parkinson’s disease (LRRK2 G2019S and GBA N370S, among the most-studied monogenic contributors)
• Hereditary thrombophilia (Factor V Leiden and prothrombin variants)
• Age-related macular degeneration
• Celiac disease
• Hereditary hemochromatosis (HFE C282Y and H63D)

A separate set of dashboard reports, including the Type 2 Diabetes polygenic risk score, is built by 23andMe Research and has not been reviewed by FDA. These are statistical, population-level scores that combine many small-effect-size variants into one number. Useful as a conversation starter, but a different category from the FDA-reviewed reports above.

Some reports are best read as polygenic risk indicators; others, like the LRRK2 Parkinson’s variant, are closer to monogenic risk flags; the research-powered scores are statistical-association estimates. Mixing those three categories together is one of the most common mistakes in consumer-DNA coverage. The dashboard does not blur them, and neither should you.

BRCA1/BRCA2 Selected Variants: The Caveat to Read Before Buying for Cancer Risk

The BRCA report is the most consequential health report 23andMe sells and the one most misunderstood. The 2018 De Novo authorization originally covered three founder variants common in people of Ashkenazi Jewish descent. The current report was expanded through 510(k) clearance K223597 (a substantial-equivalence pathway, not a new De Novo authorization) in 2023 and now checks 44 selected variants out of more than 1,000 pathogenic BRCA1 and BRCA2 variants documented in the clinical literature.

That math matters. A positive result is meaningful and warrants a genetic counselor visit. A negative result on the 23andMe BRCA report does not rule out hereditary breast or ovarian cancer risk. It tells you only that you do not carry the specific selected variants on the panel, and the listed variants are most prevalent in people of Ashkenazi Jewish descent. The FDA decision summary for K223597 is explicit on this point: the 44-variant panel does not represent the majority of BRCA1/BRCA2 variants in most ethnicities, and users with a personal or family history of cancer should talk to a healthcare professional about appropriate testing. If that describes you, a 23andMe negative result is not reassurance. Talk to a genetic counselor about clinical-grade BRCA sequencing.

The MUTYH-Associated Polyposis report (510(k) K182784, 2019) sits in the same selected-variants pattern: two MUTYH variants are checked, and absence of those two variants does not rule out other cancer-related variants, especially outside the ethnicities the panel was designed around.

Pharmacogenetics Reports

The PGx reports authorized under DEN180028 currently cover three genes per the live 23andMe reports list: CYP2C19 (3 variants, relevant to clopidogrel and certain PPIs), DPYD (2 variants, fluoropyrimidine chemotherapy toxicity), and SLCO1B1 (1 variant, statin-related side effects). The reports tell you which metabolizer category your variants suggest for those specific genes. They do not test every relevant PGx variant, they do not tell your prescriber what to prescribe, and they should not be used to start, stop, or change a medication on your own. Bring the report to a clinician and treat it as input to a conversation, not a decision.

Pricing Tiers

Three tiers, two renewal subscriptions, and a meaningful gap between what each tier unlocks. Prices verified against the 23andMe compare page on May 20, 2026; re-verify on the day you buy, since prices and promo terms have shifted multiple times since 2024.

If you cancel a Premium or Total Health subscription, you keep the reports you have already received but lose ongoing report updates and the subscription-tier features. Refund policy varies, so check the current support pages before purchasing if that matters to your decision.

Privacy, Data, and the TTAM Era

23andMe’s 2025 bankruptcy and the subsequent TTAM acquisition do not change the day-to-day product, but they do change the trust math.

The 2025 Timeline

23andMe filed for Chapter 11 bankruptcy protection on March 23, 2025. Multiple state attorneys general, including California and Connecticut, urged consumers to consider deleting their accounts and instructed the company on data-handling expectations during the sale. California’s attorney general sought to block aspects of co-founder Anne Wojcicki’s reacquisition vehicle on data-protection grounds; a federal judge declined to halt the sale, as Reuters reported in July 2025. On July 14, 2025, TTAM Research Institute, a nonprofit public-benefit corporation, completed the asset acquisition and continues to operate the consumer service. TTAM’s public-benefit-corporation charter includes a stated public-benefit purpose; that is not the same as nonprofit charitable status, but it is a stronger structural signal than a standard for-profit acquirer.

What the Current Privacy Policy Says

The legal privacy statement at /legal/privacy/ was last updated May 19, 2026; we read it on May 20, 2026. The policy separates research-program data from account-level data and details the consent choices you make at signup. Research participation is opt-in, and consent is revocable from inside the account dashboard. The statement says TTAM does not sell personal genetic data and describes the research-purpose framework under which data can be used.

Your Data-Control Options

Four meaningful levers, all documented on 23andMe’s support pages:

• Download your raw data as a tab-separated genotype file.
• Request account deletion, which removes account-linked genetic data on the platform.
• Withdraw research consent at any time, which stops future use of your data in new studies (already-shared data in completed studies cannot be unwound).
• Discard your physical sample, which prevents future re-analysis from the original saliva tube.

23andMe requires 2-step verification for direct account sign-in; Apple and Google sign-in workflows handle authentication on their side, so the 2-step prompt may not appear if you log in via those providers.

Realistic Threat Model

The risks worth weighing: future re-acquisition by a less-aligned owner, breach of the account-level dataset, law-enforcement requests for individual records, and policy-language drift over time. The data-control levers above are your best mitigation. If you are not comfortable with any consumer DNA company holding your sample, this kit is not the right choice. See our broader explainer on privacy concerns in DNA testing for the category-wide framing.

23andMe vs. AncestryDNA

If your shortlist is down to two, the choice resolves on what you actually want the test to do.

Buy 23andMe if health reporting, carrier screening, or pharmacogenetics is part of why you want a DNA test. Buy AncestryDNA if you want to build a family tree, find living relatives, or search historical records. For the full head-to-head, read 23andMe vs. Ancestry.

Alternatives Worth Considering

If 23andMe is not quite right, three alternatives cover the most common reasons buyers shop around:

• AncestryDNA: buy this instead if your real question is family history, relatives, and historical-records work.
• MyHeritage DNA: buy this instead if you want stronger international ancestry coverage, especially outside the U.S. and U.K.
• Nebula Genomics: buy this instead if you want whole-genome sequencing rather than SNP-chip genotyping. Nebula’s consumer line currently sells DNA Complete by Nebula Genomics at 1x, 30x, and 100x coverage tiers, and you should expect to pay more than a SNP-chip kit.
• ClarityX DNA: buy this instead if your only goal is pharmacogenetic testing tied to medication management.

Frequently Asked Questions

Is 23andMe worth it in 2026? Yes if you want FDA-reviewed health screening combined with ancestry. For genealogy-focused work, AncestryDNA is the stronger starting point.

Which 23andMe tier should you buy? Premium at $199 is the default if health is part of the goal. The $99 Ancestry Service tier covers ancestry only. Total Health at $499 targets buyers who want clinician-guided genetics work.

Can 23andMe diagnose a disease? No. The FDA, CDC, and 23andMe itself are consistent: results are screening tools. Confirm any health-relevant result with a clinician or genetic counselor before acting on it.

Does 23andMe test for all BRCA variants? No. The BRCA1/BRCA2 Selected Variants report checks 44 selected variants out of more than 1,000 known pathogenic BRCA variants. A negative result does not rule out hereditary cancer risk, especially outside Ashkenazi Jewish ancestry, where the listed variants are most prevalent.

Is every Premium health report FDA-reviewed? No. Carrier-screening and pharmacogenetics reports are FDA-authorized through De Novo decisions, and the BRCA1/BRCA2 Selected Variants and MUTYH reports were cleared through 510(k). Other reports in the dashboard, including the Type 2 Diabetes polygenic risk score, are powered by 23andMe Research and have not been reviewed by FDA.

Is 23andMe safe after the bankruptcy? The consumer service continues to operate under TTAM Research Institute, a nonprofit public-benefit corporation. The current privacy policy was updated May 19, 2026, and the deletion, research opt-out, and sample-discard controls remain available. The structural risk worth weighing is future re-acquisition or policy drift, not present-day operations.

The Bottom Line

We recommend 23andMe Premium if you want ancestry plus FDA-authorized carrier and pharmacogenetics screening and selected FDA-reviewed health-risk reports in one account. That is the case the product is built for, and no other mainstream kit packages those regulatory categories at this price.

Skip it if your goal is family history; start with AncestryDNA instead. If a personal or family history of cancer is the reason you are looking at DNA tests, start with a genetic counselor and clinical-grade testing, not a consumer kit. And if you want a broader health-focused shortlist before deciding, read the best DNA tests for health.