Skip to content
Book a demoStart your free trial
COI and pedigrees

Coefficient of inbreeding (COI) explained for breed clubs and registries

The short answer

A coefficient of inbreeding (COI) is the probability that an animal inherited two copies of the same gene variant from an ancestor shared by its sire and dam. The figure depends heavily on how many generations were counted and how complete those generations are, which is why two calculators rarely agree. A registry can publish COI responsibly if every figure states its method, its depth and its completeness.

Dusty Mumphrey
Founder, Breed Ledger. Senior engineer. Active multi-species breeder.

Published 9 min read

Illustration of a pedigree chart in which the lines through the sire and the dam both lead back to the same ancestor, highlighted in green, beside a percentage gauge.

This guide is for registrars, health committee members and board officers who keep hearing members quote COI figures that do not match, and who need to decide what the club should say about it. By the end you will be able to explain what the number means in a sentence, explain why two honest calculators disagree, and draft a short publishing standard the board can vote on.

What does a COI figure actually measure?

It measures probability, not a trait you can see. The Royal Kennel Club describes the coefficient as the probability that two copies of a gene variant were inherited from an ancestor common to both parents [1].

A few points help members read the number correctly:

  1. It is a percentage chance across the genome. A COI of 10% means that, on average, about one in ten gene positions is expected to carry two copies traced back to the same ancestor.
  2. Zero only means "no shared ancestors found." It does not mean none exist. It means none appear in the records the calculator looked at.
  3. Known matings give useful anchors. The Institute of Canine Biology uses 25% as the value you would expect from mating full siblings [2]. Comparing a figure to that anchor helps members picture what a number like 18% represents.
  4. It describes risk, not destiny. The Royal Kennel Club states that COI is a measure of risk and not a guarantee of health [1]. The Institute of Canine Biology describes the risk of harmful effects as rising in proportion to COI, without saying any single animal will be affected [3].

Why does the number change with the number of generations?

Because a calculator can only find shared ancestors inside the window it searches. Count five generations and every common ancestor further back is invisible.

The Institute of Canine Biology notes that a COI based on a five generation pedigree tells you about inbreeding only over those five generations [3]. In one of its worked examples, a single animal showed about 6% at five generations, about 10% at ten generations, and a value close to a full sibling mating (25%) once 30 generations were included [2]. Its answer to how many generations to use is "As many as you can" [2].

Here is how that looks for one animal in a fictional registry.

Example data: Alder Vale Registry, one dog, same pedigree file

Generations countedAncestors known in deepest generationCOI
532 of 32 (100%)3.1%
8201 of 256 (79%)6.4%
121,180 of 4,096 (29%)8.9%
All recordsVaries by line9.6%

Nothing about the dog changed between rows. Only the window did. The last two rows also show the second problem: gaps in the record.

Why do two calculators give different answers for the same animal?

Usually because they are working from different records, different depths, or both. Neither has to be wrong.

The common causes are:

  1. Different generation limits. The Institute of Canine Biology observes that some people use 5 generations, some 8 and some 10 [2]. A ten generation figure and an all records figure are not comparable.
  2. Different databases. Each online calculator works from its own database. If one site holds a line back to the foundation stock and another stops at an import, they will disagree.
  3. Missing ancestors. When relevant ancestors are absent from the pedigree, the true level of inbreeding is underestimated [3]. Researchers studying Dalmatians reached the same conclusion from data: missing family members and the assumption that founders are unrelated bias pedigree COI toward lower values [4].
  4. Imports with short pedigrees. The Royal Kennel Club warns that for imported dogs it may hold only a three generation pedigree, so the result may be less informative, and it tells users to check how many generations were used [1].
Tip

When two members disagree, ask each one three questions before debating the number: which calculator, how many generations, and how complete was the pedigree. Most arguments end there.

Should the club watch individual animals or the breed as a whole?

Both, but the breed level figures are the club's job. Individual COIs help breeders plan matings. Breed average COI and effective population size tell the board whether the whole population is losing diversity.

Breed average COI. The Royal Kennel Club calculates COI for every dog in a breed each June, then averages the values for dogs born in the UK and registered in the previous year [1]. It advises breeders to aim for a COI at or below that breed average, and ideally as low as possible [1]. A published average gives members a shared reference point, and a yearly series shows the direction of travel.

Effective population size (Ne). This is the size of an idealised population that would gain inbreeding at the same rate per generation as the real one [5]. It can be far smaller than the number of animals registered. A study of all UK Kennel Club breeds found no statistically significant link between Ne and census population size measured by registrations [5]. The same paper cites published guidance that diversity loss speeds up sharply when Ne falls below 100, and that a population with Ne under 50 is considered at high risk [5].

For a board, the practical point is that a breed with thousands of registrations can still have a small Ne if a few popular sires dominate. The authors also caution that incomplete pedigrees for imported animals can lead to underestimates of the inbreeding rate and Ne [5].

What can a registry responsibly publish?

A figure with its working shown. A bare percentage invites the arguments described above; a labelled one ends them.

A short publishing standard might require every COI the registry releases to state:

  1. Method. Pedigree calculation from registry records, or a genomic estimate from a DNA panel, and which one.
  2. Generations used. Either a fixed number or "all available records," applied the same way every time. The Royal Kennel Club, for example, does not limit the number of generations [1].
  3. Completeness. The share of ancestors known in each generation, or at least a flag when the pedigree is shorter than the standard depth.
  4. Date calculated. COI can rise as older records are added, so a figure needs a date.
  5. Breed context. The current breed average calculated the same way, so a member can see whether an animal sits above or below it.

Whatever depth the board picks, using it consistently matters more than the exact number, because a figure calculated the same way each year can be compared with last year's.

Warning

Avoid publishing COI on its own as a ranking or a pass mark. It is a probability about risk, not a health result, and a low figure built on a thin pedigree can look better than it is.

What are the limits, and where do DNA measures fit?

COI is not a health test, and pedigree COI is only as good as the records behind it. DNA based measures of inbreeding exist and can fill some of the gaps.

Genomic estimates look directly at stretches of the genome where both copies match, often called runs of homozygosity. In the Dalmatian study, pedigree and genomic inbreeding measures were highly correlated, yet the authors judged the genomic figure more accurate because pedigree figures are pulled down by missing ancestors [4]. Genomic figures have their own assumptions: the Institute of Canine Biology notes that choosing different segment lengths gives different estimates, so a DNA COI is also an estimate [3].

For a club, the two approaches work best together. Pedigree COI is cheap and covers every registered animal. Genomic estimates are useful for checking lines where the pedigree is short, such as imports or foundation stock. Neither replaces breed specific health screening.

If the board is weighing new software for the registry's records, the registries page describes what Breed Ledger offers registries.

Key takeaways

  • COI is the probability that an animal carries two copies of a gene variant from a shared ancestor; it describes risk, not health.
  • The figure grows as more generations are counted and shrinks when ancestors are missing, so depth and completeness must travel with the number.
  • Calculators disagree mainly because they use different records and different generation limits.
  • Breed average COI and effective population size are the board level measures worth tracking every year.
  • Publish every COI with its method, generations used, completeness and date.

Frequently asked questions

Is a lower COI always better?

Lower is generally less risky, and the Royal Kennel Club advises breeding at or below the breed average [1]. A low figure from a short or patchy pedigree may be an underestimate, so check completeness first.

Which calculator should the club endorse?

Endorse a method rather than a website: the records used, the generation depth and how completeness is reported. Any calculator that meets that standard and shows its working will produce comparable figures.

Can a DNA test replace pedigree COI?

It can add to it. Genomic estimates avoid the missing ancestor problem but rely on their own assumptions [3]. One practical mix is pedigree COI for every animal plus genomic checks where pedigrees are short.

How often should a breed average be updated?

Once a year is a common rhythm. The Royal Kennel Club recalculates its breed averages every June using the previous year's UK born registrations [1].

Sources

  1. The Royal Kennel Club, "Inbreeding calculators" (undated), https://www.royalkennelclub.com/health-and-dog-care/health/health-testing-and-screening/inbreeding-calculators (external site), checked 2026-11-09.
  2. The Institute of Canine Biology, "How many generations of pedigree data should you use to estimate inbreeding?" (2014-08-20), https://www.instituteofcaninebiology.org/blog/how-many-generations-of-pedigree-data-should-you-use-to-estimate-inbreeding (external site), checked 2026-11-09.
  3. The Institute of Canine Biology, "10 key things to know about the coefficient of inbreeding" (2024-12-17), https://www.instituteofcaninebiology.org/blog/10-key-things-to-know-about-the-coefficient-of-inbreeding (external site), checked 2026-11-09.
  4. Canine Medicine and Genetics, "Demographic assessment of the Dalmatian dog: effective population size, linkage disequilibrium and inbreeding coefficients" (2020), https://pmc.ncbi.nlm.nih.gov/articles/PMC7371805/ (external site), checked 2026-11-09.
  5. Canine Genetics and Epidemiology, "Trends in genetic diversity for all Kennel Club registered pedigree dog breeds" (2015), https://pmc.ncbi.nlm.nih.gov/articles/PMC4579366/ (external site), checked 2026-11-09.
Share this articleLinkedIn (external site)Email

About the author

Dusty Mumphrey

Written by

Dusty Mumphrey

Founder, Breed Ledger. Senior engineer. Active multi-species breeder.

Dusty builds Breed Ledger and breeds more than one species himself. He writes about why club records matter and what he is building next.

About Breed Ledger

Registries

See how Breed Ledger works for registries.

Registrations, certificates and records, run from one place.