Guides › Donor matching

What 8/8 and 10/10 HLA match grades mean, and how they are counted

Updated 2026-10-29 · allele examples checked against IPD-IMGT/HLA 3.66.0 on 2026-10-29

An HLA match grade such as 8/8 or 10/10 counts how many of a donor's and recipient's alleles agree across a fixed set of HLA loci. Each locus contributes two alleles, one per chromosome, so an 8/8 grade covers four loci and a 10/10 grade covers five.

The numbers look simple, and that is the risk. Getting them right depends on which loci are counted, at what resolution, how homozygous loci are handled, and what happens when an allele is not expressed. This guide explains the arithmetic. It is not guidance on donor selection, which is a clinical decision governed by the transplant centre's own criteria.

Which loci each grade counts

The 8/8 definition is the one used, for example, by Lee and colleagues in their 2007 Blood study of 3,857 unrelated-donor transplants: high-resolution matching for HLA-A, -B, -C and -DRB1. The table above is how HLA-Verify's published counting rules define each framework. Check which definition a report uses before comparing grades across sources.

Count per chromosome, not per locus

Each locus has two alleles, so each locus is worth two points. A locus where one allele matches and the other does not scores 1 of 2, not 0 of 1.

At allele level, the comparison is between the two-field names. If a recipient is B*07:02 and B*44:02, and a donor is B*07:02 and B*44:03, the B locus contributes one match and one mismatch, and with A, C and DRB1 identical the result is 7/8.

A common mistake, and one that language models make in the HLA-Bench matching tasks, is to count matched loci instead of matched alleles: calling the example above "3 of 4 loci matched". Every grade in this family is a count of alleles.

Direction: GvH and HvG

A mismatch can matter in one direction and not the other. This happens when one side is homozygous at a locus.

HLA-Verify's rules count:

• HvG (host-versus-graft) mismatches: donor alleles absent from the recipient.
• GvH (graft-versus-host) mismatches: recipient alleles absent from the donor.

Checked against IPD-IMGT/HLA release 3.66.0: a recipient typed B*07:02, B*44:02 and a donor typed B*07:02, B*07:02 (homozygous) gives 7/8 at an 8/8 framework, with one GvH mismatch and zero HvG mismatches. The donor has nothing the recipient lacks; the recipient has B*44:02, which the donor lacks.

A single grade hides this. If direction matters for the question you are answering, report both counts.

The null-allele trap

Some alleles are not expressed at all; they carry the suffix N. A null allele can look like a match on paper because it shares its group with expressed alleles. A*24:09N belongs to the G group A*24:02:01G, and a typing reported only as A*24 covers both.

Checked against release 3.66.0, a recipient with A*02:01, A*24:02 and a donor with A*02:01, A*24:09N, identical elsewhere, gives 7/8: the A locus is a mismatch, and the result carries a null_allele flag. Treating A*24:09N as a match for A*24:02 because both are "A24" would report 8/8 for a donor whose second A allele is not expressed.

The expression suffixes guide explains N and the other suffixes.

When the typing is not precise enough

A grade at allele level needs allele-level typing on both sides. If one side is typed only to the first field, for example A*02, the A locus cannot be called a match or a mismatch.

There are two honest ways to handle that: say the locus is unresolved, or resolve the typing further. The dishonest way is to count it either way. In HLA-Verify, a locus whose typing is too coarse for the framework is reported as potential, flagged resolution_insufficient, and left out of the denominator. With the donor's A typed as A*02 and the other loci identical, the result is reported over the three resolvable loci, 6/6, with the A locus marked potential, rather than as 8/8.

Read a grade together with its denominator and flags. "6/6 with one locus potential" and "8/8" are not the same statement. The typing resolution guide covers what each level can resolve.

Names have to agree before counts can

Match counts compare names, so every name has to be valid and current in the same release first:

• Legacy spellings such as Cw*07:02 should be read as C*07:02, not counted as a mismatch.
• Deleted names should be followed to their successor before comparison.
• Group names such as A*02:01:01G compare by the two-field name they are written under, and their member lists change by release.
• Multiple allele codes cannot be compared without expanding them, so a locus reported only as a code is unresolved.

A grade computed over a stale or misspelled name is wrong even if the arithmetic is right.

What match grades do not tell you

A match grade is a count over a defined set of loci. It does not include loci outside the framework, such as DQA1, DPA1 or DPB1 in an 8/8 grade, and it says nothing about antibodies. In the Lee 2007 cohort, mismatching at HLA-DP or -DQ was not associated with survival, but that is one study's finding about outcomes, not a property of the counting. HLA-Verify's matching rules state that DPB1 T-cell-epitope permissiveness is out of scope for these frameworks.

Frequently asked questions

What loci are in an 8/8 HLA match?

HLA-A, -B, -C and -DRB1, at allele level, two alleles each. This is the definition used by Lee and colleagues in their 2007 Blood study.

What does 10/10 add to 8/8?

HLA-DQB1. A 10/10 grade counts HLA-A, -B, -C, -DRB1 and -DQB1 at allele level.

Is a 7/8 match one mismatched locus?

It is one mismatched allele out of eight. That allele sits at one locus, but the grade counts alleles, not loci, so two mismatches at one locus would make it 6/8.

Why does a homozygous donor change the count?

Because direction matters. A donor homozygous for an allele the recipient also carries has no allele the recipient lacks (no HvG mismatch), while the recipient's other allele is absent from the donor (one GvH mismatch).