How to find your blood type in 23andMe raw data

A curious genome detour · Updated June 2026

23andMe & raw DNA

Why 23andMe doesn't just tell you your blood type

Blood type is genetic, so you'd think it would be a one-line report. The catch is that 23andMe genotypes specific positions, and accurate ABO typing depends on reading the exact combination of those positions correctly. Rather than risk a confident-but-wrong answer, 23andMe leaves blood type out of its reports. The information is still in the raw data, though — you just have to look it up yourself.

What ABO blood type actually is

Before decoding anything, it helps to know what you're decoding. Your blood type describes the antigens — little sugar structures — sitting on the surface of your red blood cells. In the ABO system there are two possible antigens, cleverly named A and B. Whether you build them, and which ones, is what makes you type A, B, AB, or O.

Here's the mechanism, because it's genuinely elegant. The ABO gene, on chromosome 9, codes for an enzyme that sticks a specific sugar onto a base molecule already present on your red cells (the H antigen). There are three main versions of the gene:

• The A version builds an enzyme that adds one particular sugar, creating the A antigen.
• The B version builds a slightly different enzyme that adds a different sugar, creating the B antigen. The A and B enzymes differ by only a handful of amino acids, but that's enough to change which sugar they attach.
• The O version is a loss-of-function copy. Somewhere along the line it picked up a mutation that breaks the enzyme, so it adds nothing. People with two O copies just keep the plain H antigen — no A, no B. That "makes nothing" quality is the key to reading O from your DNA.

ABO genotype-to-phenotype in 30 seconds

You inherit one ABO allele from each parent, and they combine like this:

• A = AA or AO
• B = BB or BO
• AB = one A and one B — both enzymes work, so you build both antigens (A and B are codominant)
• O = OO — both copies are broken, so you build neither

Notice that A and B "win" over O. An AO person is blood type A because their one working A gene still builds the A antigen; the broken O copy just sits there quietly. That's why O is described as recessive — it only shows up as type O when there's nothing else doing the work.

The trick to reading all this from raw data is that the O allele is mostly defined by a tiny deletion that disables the enzyme, while A versus B comes down to a different marker that changes which sugar the working enzyme attaches.

The key markers in your 23andMe raw data

Two SNPs do most of the work for ABO:

• rs8176719 — this is the famous single-base deletion that creates the O allele. In 23andMe files it usually shows up as an insertion/deletion. A genotype carrying the deletion (often written as D, -, or a missing base) points to an O allele on that chromosome; a present base (a G insertion) points to a non-O (A or B) allele.
• rs8176746 — this distinguishes A from B. Broadly, the G base is associated with the A allele and the T base with the B allele. A related marker, rs8176747, is sometimes used the same way to firm up the A/B call.

Putting them together: rs8176719 tells you how many O alleles you likely have, and rs8176746 tells you whether your non-O allele(s) are A or B. From there you can reconstruct A, B, AB, or O.

What about the + and − (Rh status)?

The plus or minus after your blood type comes from a completely separate system, the Rh system, driven mainly by a different gene: RHD. The "D" in RhD refers to the D antigen. If you make it, you're Rh-positive; if you don't, you're Rh-negative. So "A positive" really means "type A in the ABO system, and you carry the D antigen."

Here's why Rh is harder to read than ABO. Rh-negative most commonly happens when the entire RHD gene is simply deleted — it's not there to make the antigen. And you can't directly genotype a gene that's missing; there's no sequence to read. So instead of a clean single-SNP answer, 23andMe relies on nearby tag SNPs in the Rhesus region that tend to travel with the deletion. Two copies of the deletion-tagging pattern suggest Rh-negative; otherwise you're most likely Rh-positive.

Because it's an indirect correlation rather than a direct readout — and because the real biology of Rh is more complicated than a simple present/absent switch, with rarer variants and partial antigens — Rh is genuinely the softest part of the estimate. If any part of a DNA-inferred blood type is going to be wrong, it's most likely to be here.

Where the inference gets uncertain

This is the honest core of the whole exercise. Mapping genotype to blood type is clean and reliable for many people, but not everyone:

• Encoding quirks. Deletions (the heart of the O allele) are messy in consumer files. The same deletion can appear as --, DD, DI, II, or simply be absent, depending on the file version and reference build. Misreading one row flips your answer.
• Rare alleles. There are less common ABO subtypes and weak variants that the handful of tested SNPs don't distinguish.
• Rh's missing gene. You're reading tag SNPs that correlate with the RHD deletion, not the deletion itself, so linkage can occasionally break down.
• No-calls. If the exact position wasn't read well on your chip, you get a blank, and part of the puzzle is simply missing.

None of this makes the exercise pointless — it's a fun and usually-accurate reconstruction. It just means the right mental model is "well-informed guess," not "test result."

How to look up these SNPs in your file

First you need the file itself. If you haven't downloaded it yet, follow our guide to downloading your 23andMe raw data. Then:

1. Open the .txt file in any text editor, or load it into a tool that can search by rsid.
2. Search for rs8176719 and note the genotype in the last column.
3. Search for rs8176746 (and rs8176747 if present) and note those genotypes.
4. Search the RHD-region markers for your Rh estimate.
5. Combine them: O-deletion count → how many O alleles; A/B marker → which non-O allele; RHD markers → + or −.

For example, two copies of the O deletion at rs8176719 point toward type O, while one O allele plus an A-associated rs8176746 points toward type A. It's satisfying when the pieces line up — but remember it's a reconstruction, not a lab result.

If hunting through a text file sounds tedious, our browser-based DNA viewer lets you search these rsids in your own file without uploading it anywhere, which makes the encoding quirks easier to spot.

Who can donate to whom (the compatibility basics)

Once you know your ABO and Rh type, the famous donor chart makes a lot more sense. The rule is simple: your immune system attacks antigens it doesn't recognize. So a type A person makes antibodies against B, a type B person makes antibodies against A, type O makes antibodies against both, and type AB makes antibodies against neither.

That produces two well-known extremes for red-cell transfusions:

• O-negative is the classic universal donor — no A, no B, and no D antigen means there's nothing on those cells for a recipient's immune system to flag.
• AB-positive is the universal recipient — you already have A, B, and D, so foreign red cells carrying any of them don't trigger an attack.

This is nice intuition to have, but note the caveat: real transfusion medicine involves far more than ABO and Rh (dozens of other blood group systems, plus a cross-match step), which is exactly why it's done in a lab and never from a DNA guess.

A fun population footnote

Blood types aren't evenly distributed around the world. O and A are common in many populations, B is generally less common, and AB is the rarest of the four. Rh-negative is a minority pattern globally but noticeably more frequent in some European-descended populations than in most East Asian ones. So the specific mix of A, B, AB, O, and +/− is a little fingerprint of ancestry and migration — which is part of why blood type shows up in population genetics at all. We're keeping this qualitative on purpose: any single "X% of people are type O" figure hides just how much it swings by region.

What about the blood-type-and-health claims?

You've probably seen headlines linking blood type to heart disease, certain infections, or even the infamous "blood type diet." Here's the grounded version: researchers have found some weak statistical associations between ABO type and a handful of conditions — the ABO gene influences a few molecules in the blood beyond the red-cell antigens, so it's biologically plausible that it nudges risk here and there. But these effects are small, incompletely understood, and not actionable. Knowing your ABO type won't change what you should eat, how you should train, or what screening you need. The "blood type diet" in particular has no solid evidence behind it. Treat blood type as an interesting fact about your biology, not a health lever.

The grounded takeaway

Your 23andMe or AncestryDNA raw file usually carries enough information to make a solid guess at your ABO type and a shakier guess at your Rh status. It's a satisfying detour — you get to see the actual molecular machinery (a sugar-adding enzyme, a present-or-missing gene) that produces something as everyday as "A positive." But the whole exercise lives firmly in the "curious, not clinical" category. Genotype inference can be wrong, it's not calibrated for medical use, and a real serological test is cheap, fast, and definitive. Enjoy the puzzle, confirm with a lab if it ever matters, and don't let a text file talk you out of a blood test.

Want the bigger picture of everything your raw file can reveal? Start with our complete guide to 23andMe raw data.

Frequently asked questions

Is 23andMe blood type accurate?

It's an estimate, not a guarantee. The ABO markers in your raw data give a good prediction for most people, but genotype encoding quirks and the indirect way Rh status is inferred mean it can be wrong. Always confirm with a real lab blood test before relying on it for anything.

Which SNPs show blood type in 23andMe raw data?

The main ones are rs8176719 (the deletion linked to the O allele) and rs8176746 (A versus B), sometimes with rs8176747 as backup. Rh status is inferred from RHD-region markers rather than a single clean SNP.

Can I find my Rh (+/−) status too?

Roughly. Rh-negative is usually caused by a full deletion of the RHD gene, which can't be genotyped directly, so 23andMe relies on nearby tag SNPs that correlate with it. That makes Rh the least certain part of the estimate — a lab test is the only way to be sure.

Can I use my inferred blood type for a transfusion or emergency?

No — never. A DNA-inferred blood type is a genotype prediction, not clinical-grade typing, and it can be wrong. Transfusions, surgery, pregnancy and Rh management, and emergencies all use a proper serological blood test performed by a lab. If your blood type actually matters, get tested.

What actually makes A, B, AB, and O different?

The ABO gene codes for an enzyme that adds a sugar onto a base molecule on your red blood cells. The A version of the gene builds the A antigen, the B version builds the B antigen, and the O version is a broken (loss-of-function) copy that builds neither. AB means you inherited one working A gene and one working B gene; O means both copies are non-functional.

Why is O the most common blood type?

O is common across many populations, but the exact distribution of A, B, AB, and O varies a lot by ancestry and region. Because O comes from loss-of-function versions of the gene that arose and spread widely, it's frequent almost everywhere — but there's no single global percentage that applies to everyone.

Does my blood type affect my health?

Research has found weak statistical associations between ABO type and a few conditions, but the effects are small, not fully understood, and not something to act on. Your blood type isn't a meaningful health lever the way diet, sleep, or exercise are. Treat any ABO 'health' claim with heavy skepticism.

Why doesn't the app just show me my blood type directly?

Because doing it responsibly means being honest about uncertainty. The genotype-to-phenotype mapping is clean for many people but genuinely ambiguous for others, especially on the Rh side. Showing a confident single answer would imply a reliability that inferred typing doesn't have.