Skip to content
Available PuppiesPlanned LittersAvailable DogsVeterans Discount
Peterbuilt Boerboels

COI and Breeding

The coefficient of inbreeding is the most quoted number in dog breeding and the most casually misused. Here is what it actually measures, why the same dog can honestly be given two very different COIs, what the research says rising inbreeding costs, and the part almost nobody knows: no major registry sets a limit on it.

The short answer

COI is a probability, not a verdict.It estimates how much of a dog’s genome is expected to be doubled up because the same ancestor appears on both sides of the pedigree. Under 5% is the usual target, 5 to 10% carries modest detrimental effects, and above 10% the effects become significant.[1]

The number is worthless without its generation depth. A COI calculated over five generations and one calculated over ten are not comparable, and the shallow one is always the flattering one. If a breeder quotes you a COI and cannot tell you how many generations it came from, you have been handed a marketing figure.[1]

And no major registry polices it. We read the current SABBS Constitution, Bylaws and Code of Conduct, plus every archived edition back to 2015, and the AKC registration rulebook. None of them contain a coefficient-of-inbreeding limit, and none refuse a litter for being too closely bred.[7, 8, 9]The UK’s Royal Kennel Club is the strictest of the three and it still only bans the three first-degree pairings.[6] Keeping COI low is a standard a breeder chooses to hold, not a rule anyone enforces on them.

On this page

  1. The short answer
  2. What COI actually measures
  3. A COI without its generation depth is not a number
  4. Pedigree COI vs measured DNA
  5. What actually goes wrong as COI rises
  6. What the registries actually require
  7. Where the Boerboel sits
  8. Is linebreeding always a mistake?
  9. How to read a COI you are handed
  10. Frequently asked questions
  11. Sources & further reading

What COI actually measures

Every dog carries two copies of each gene, one from each parent. The coefficient of inbreeding (COI, or F, from Sewall Wright’s 1922 formulation) is the probability that those two copies are identical by descent: not merely similar, but literal copies of one ancestral gene that reached the puppy down two separate paths in the pedigree.[1]

In plainer terms, it is the share of the genome you would expect to be doubled up. A dog with a COI of 10% is expected to be homozygous by descent at roughly a tenth of its genome. That matters because recessive problems only surface when a dog inherits two broken copies. Doubling up is also exactly how a breeder fixes a trait they want, which is why inbreeding is a tool and not simply a sin. It concentrates whatever is there, good and bad, without discriminating.

Here is what standard matings produce, assuming the two parents are themselves completely unrelated and uninbred:

25%

Full siblings, or parent to offspring (father x daughter)

12.5%

Half-siblings, or grandparent to grandchild

6.25%

First cousins

3.125%

First cousins once removed

Reference COI values for common matings.[1]

The catch that trips people up.Those textbook figures assume unrelated parents. In a closed breed nobody’s parents are truly unrelated, so the real coefficient of a given mating is the textbook figure plus whatever both parents already carried. A half-sibling mating in a tight breed does not produce a 12.5% puppy. It produces something meaningfully higher, and only a full pedigree calculation will tell you how much higher.

A COI without its generation depth is not a number

This is the single most important thing on this page, and it is the thing most often left out of a sales conversation.

A COI is not a property of a dog. It is the output of a calculation run over a particular slice of pedigree. Run it over five generations and the software can only find duplicated ancestors within those five generations. Every shared ancestor sitting in generation six, eight or twelve is invisible, and the number comes back low. Run the same dog over ten or twenty generations and all that hidden overlap is suddenly counted, and the number climbs.[1]

The Institute of Canine Biology is blunt about the consequence: a five-generation coefficient “will be an estimate of the probability of inheriting two copies of the same allele from only the animals in those 5 generations.” Their guidance is to use at least eight to ten generations, and twenty would be better, ideally against a database that reaches back to the breed’s first registered dogs.[1]

What this means in practice

The same puppy can be advertised, entirely honestly, as having a COI of 2% or a COI of 11%. Neither figure is a lie. One was run shallow and one was run deep. This is why comparing a COI from one breeder against a COI from another is meaningless unless both were calculated to the same depth, on the same database. It is also why the shallow figure is the one that tends to appear in advertising.

So the first question to ask about any COI is never “is that good?” It is “over how many generations, and on which database?” A breeder who works with these numbers will answer instantly. If the answer is vague, the number should be treated as decoration.

Pedigree COI vs measured DNA

There are two entirely different ways to get a coefficient of inbreeding, and they do not agree.

A pedigree COI is arithmetic run on a studbook. It is an expectation, and it rests on an assumption that is almost never true in a closed breed: that the founders at the top of the tree were unrelated to one another. They rarely were. Every relationship that predates the studbook is simply invisible to the calculation, so the result is systematically optimistic.[11]

A genomic COI (usually reported as FROH, from runs of homozygosity) reads the DNA the dog actually inherited. It does not care what the paperwork says. It also captures the luck of the draw: two full siblings have the same pedigree COI of 25%, but their measured genomic inbreeding can differ, because meiosis does not deal the same hand twice.

How far apart do they land? The Bullmastiff is the useful comparison, because both kinds of estimate exist for it. Its pedigree effective population size was put at roughly 41, but measuring actual DNA dropped the molecular figure to about 29, with roughly 16% of the genome sitting in long doubled-up runs.[12] The direction of the error is consistent and worth remembering: the pedigree number is the optimistic version.

The practical upshot.If a breeder shows you a pedigree COI, read it as a floor rather than a measurement. If they show you a genomic COI from a DNA panel, that is a stronger claim, but check it is actually a genomic figure and not a pedigree calculation displayed inside a DNA company’s dashboard. The two get confused constantly.

What actually goes wrong as COI rises

“Inbreeding depression” is the umbrella term, and it is worth being specific about what has actually been measured in dogs rather than asserted.

Smaller litters, and fewer puppies surviving

Leroy and colleagues examined litter size and survival across seven French breeds. Litters with a coefficient above 12.5% averaged about 0.8 fewer puppies than litters below 6.25%, and inbreeding showed a negative effect on individual survival as well as on litter size.[3] For a breed with modest litters to begin with, most of a puppy per litter is not a rounding error.

The same result from DNA, not pedigrees

The obvious objection to pedigree studies is that pedigrees are unreliable. So Chu and colleagues repeated the question using measured genomic inbreeding in 93 reproductively intact female Golden Retrievers, and found the same direction: higher FROH was associated with lower litter size.[4] When the pedigree evidence and the DNA evidence point the same way, the finding is hard to dismiss.

More trips to the vet, across the board

Bannasch and colleagues combined genotype-based inbreeding with body weight and insurance morbidity data across 227 breeds. Average inbreeding came out at Fadj = 0.249, which is to say most breeds are inbred at a level well past anything a careful breeder would plan. Breeds in the 0.126 to 0.25 band carried a 22% higher rate of veterinary care events than the least inbred breeds, and breeds above 0.25 carried 29% higher.[5]

One honest caveat: that study compares breeds, not individual dogs. It tells you that heavily inbred populations cost more in veterinary care. It does not let you predict a specific puppy’s vet bills from its personal coefficient.

Put together, the picture is consistent rather than apocalyptic. Rising inbreeding does not doom a litter; it shifts the odds, and it shifts them first and most measurably in reproduction. The effects are cumulative across generations, which is the real reason to care. A single moderately close mating is a decision. A kennel that makes that decision every generation is running a trend, and trends in closed populations only move one way unless someone deliberately pushes back.

What the registries actually require

Buyers routinely assume that a registry somewhere is checking this. For the two registries that matter most to a Boerboel buyer, it is not.

We read the SABBS Constitution, Bylaws and Code of Conductin their current editions, both accepted on 26 June 2026, along with every archived edition back to the original 2015 bylaws and the predecessor SABT documents. The words “inbreeding” and “coefficient of inbreeding” do not appear in any of them.[8, 9] Bylaw E, which governs matings, restricts age (no breeding under twelve months), appraisal status (unappraised dogs cannot be bred), minimum appraisal score, and breeding to dogs from non-conforming registries. It says nothing whatever about breeding related dogs.[8]

We also read the AKC’s Rules Applying to Registration and Discipline. It contains no provision on inbreeding, coefficients, or the relatedness of the parents.[7] The AKC can refuse a registration over unverified parentage, falsified information or failure to keep required records. How closely the sire and dam are related is not on the list.

RegistryCOI limitFirst-degree matingsWhat the rules actually say
SABBS (South African Boerboel Breeders' Society)NonePermittedBylaw E (Matings) restricts age, appraisal status and cross-registry breeding. It says nothing about related dogs. The word inbreeding does not appear in the Constitution, the Bylaws or the Code of Conduct.
AKC (American Kennel Club)NonePermittedThe registration rulebook contains no provision on inbreeding, coefficients of inbreeding, or how closely related the parents may be.
Royal Kennel Club (UK)None, but see next columnRefusedThe RKC will not register a litter from a father-daughter, mother-son or brother-sister mating, save in exceptional circumstances or for scientifically proven welfare reasons. There is still no numeric COI ceiling.

The Royal Kennel Clubin the UK is the outlier, and the contrast is instructive. It refuses to register a litter where “the offspring are the result of any mating between father and daughter, mother and son or brother and sister, save in exceptional circumstances or for scientifically proven welfare reasons.”[6] That is a real, enforced restriction. Note what it is not: it bans three specific pairings, not a coefficient. A breeder can repeat half-sibling and cousin matings generation after generation, drive the true coefficient well past anything a father-daughter cross would have produced, and remain entirely within the rules.

The one place SABBS touches the subject. The SA Stud Book Logixsystem that SABBS breeders use includes an inbreeding calculator, and the 2015 breeder guide notes that Logix “will show it in red if it is above 6%.”[10] That is the closest thing to a published threshold in the entire SABBS ecosystem, and it is a colour in a third-party herd-book tool. No bylaw references it, no registration depends on it, and nothing happens if a breeder proceeds anyway.

The closest thing to a disciplinary hook is the Breeders’ Code of Conduct, which requires selecting breeding stock “while conforming to the Society’s Breed Standard and the genetic principles of breeding at all times.”[8] Breaching the Code is misconduct under the Constitution.[9] But that is a discretionary catch-all with no number attached, and a breeder challenged under it could fairly point out that the Society publishes no limit to have exceeded.

None of this is an argument that the registries are negligent. Appraisal-based registries like SABBS are built to judge the dog in front of them, and the health gates they do enforce, hips and elbows, vaginal hyperplasia grading, eyelid assessment, are phenotypic screens of real value.[16]It is simply worth knowing where the floor is. On inbreeding, the floor is the breeder’s own judgement, and nothing else.

Where the Boerboel sits

The one peer-reviewed genetics study of the breed analysed 87,755 studbook records for dogs born between 1971 and 2019. It found that 91.2% of registered Boerboels carry some inbreeding, that mean inbreeding among those dogs is 7.5%, and that the breed-wide average rose from about 4.7% in 1983 to about 9.1% by 2019. Roughly ten ancestors account for half the entire gene pool, and effective population size sits at about 83.[2]

Two things follow. First, the study’s own authors describe the breed’s inbreeding as high, exceeding the 5% limit they cite from the UN’s Food and Agriculture Organization, so 7.5% should not be waved off as moderate.[2] Second, all of that is pedigree arithmetic, which means it is the optimistic version. No genomic coefficient of inbreeding has ever been published for the breed, and the real figure is very likely higher.[2, 11]

An effective population size around 83 is not unusual. In a study of 116 UK Kennel Club breeds it ranged from about 24 to over 900, and 68 of the 116 came in under 100.[13] The Royal Kennel Club treats above 100 as broadly sustainable and below 50 as high risk, which puts the Boerboel in a caution zone rather than a crisis.[14] The honest summary is the one the researchers themselves reached: not endangered, but genuinely in need of active management.[2]

Which brings the two halves of this page together. The breed is accumulating inbreeding, the registry sets no limit on it, and so the trend is decided entirely by the sum of individual breeders’ choices. There is a fuller treatment of the founder bottleneck and the breed’s genetics on our Boerboel genetics page.

Is linebreeding always a mistake?

No, and it is worth saying so plainly, because the opposite position is just as lazy as ignoring COI altogether.

Linebreeding and inbreeding are the same genetic process. There is no threshold at which one becomes the other, and the coefficient does not care which word is used in the advert. What differs is intensity and intent. Deliberately concentrating on an exceptional, well-tested ancestor is how consistency gets built into a line, and it is how every established breed was made in the first place.

The problem is not the tool. It is three specific failures of how the tool gets used:

  • Concentrating on a dog nobody has tested. Doubling up on an ancestor amplifies whatever that ancestor carried. If the dog has no health testing behind it, the breeder is amplifying an unknown.
  • Never coming back out. A close mating followed by deliberate outcrossing is a strategy. A close mating every generation is a ratchet, and the coefficient only climbs.
  • The popular sire effect.The fastest way to wreck a breed’s diversity is not one breeder doing one tight mating. It is a hundred breeders all using the same fashionable stud, which quietly makes everyone’s future matings closer than they look. Ten ancestors accounting for half the Boerboel gene pool is exactly that effect, recorded.[2]

So the useful question about a planned mating is never just the number. It is: what is being concentrated, is it actually proven worth concentrating, and what is the plan for the generation after this one?

How to read a COI you are handed

If you are buying a puppy and a breeder quotes a coefficient, these are the questions that separate a real figure from a decorative one. None of them are hostile, and a breeder who works with these numbers will enjoy being asked.

Over how many generations was it calculated?

The one question that makes the number mean anything. Eight to ten generations is a reasonable working minimum; five is flattering and close to uninformative.

Is it a pedigree COI or a genomic one?

A pedigree figure is a floor. A genomic figure from a DNA panel is a measurement. Both are useful, but they are different claims and should not be presented interchangeably.

Which database was it run against?

A calculation is only as deep as the records behind it. A figure from a registry database that reaches back to the founding dogs of the breed means more than one from a partial private pedigree.

What is being concentrated, and what is its health testing?

If the pedigree doubles up on a particular ancestor, ask what that dog was tested for and what it scored. Concentrating an untested dog is where the real risk lives.

What is the plan for the next generation?

A breeder managing inbreeding deliberately can tell you where the outcross is coming from. A breeder who has never thought about it will treat the question as strange.

And keep it in proportion. COI is one input, not a grade. A puppy from a 6% pairing of two thoroughly health-tested parents with sound temperament is a better bet than a 2% puppy out of two untested dogs. The coefficient tells you about risk concentration; it tells you nothing about what is being concentrated. Read it alongside hip and elbow scores, the rest of the health panel, and the temperament of the actual parents.[16]

Frequently asked questions

What is a good COI for a puppy?

Under 5% is the target most geneticists give, with 5 to 10% carrying modest detrimental effects and above 10% carrying significant ones. But those bands only mean something if the figure was calculated over a deep pedigree. A 5-generation COI of 3% and a 10-generation COI of 3% are not the same claim, and the shallow one is usually the more flattering. Always ask how many generations the number came from before you judge it.

Does SABBS penalize breeders for a high COI?

No. There is no coefficient-of-inbreeding limit anywhere in the SABBS Constitution, Bylaws or Code of Conduct, including the current editions accepted on 26 June 2026. Bylaw E, which governs matings, does not restrict breeding related dogs at all, and a father-to-daughter litter is fully registerable provided both parents are appraised, scored, DNA-profiled and microchipped. The only inbreeding content SABBS publishes is the SA Stud Book Logix calculator, which shows a planned mating in red above 6%. That is a display convention in a herd-book tool, not a rule, and nothing follows from crossing it.

Will the AKC refuse to register an inbred litter?

Not on inbreeding grounds. The AKC's registration rulebook has no provision covering inbreeding, coefficients of inbreeding, or how closely the parents are related. Registration can be refused for things like unverified parentage, falsified records or failure to keep required records, but a parent-to-offspring mating is not among them.

Is any registry stricter?

Yes. The UK's Royal Kennel Club refuses to register litters bred from father-daughter, mother-son or brother-sister matings, except in exceptional circumstances or for scientifically proven welfare reasons. Even there the rule bans the three first-degree pairings rather than setting a numeric COI ceiling, so a breeder can still stack repeated half-sibling and cousin matings and stay inside the rules.

What actually happens when COI goes up?

The best-documented costs are reproductive. Across seven French breeds, litters with a coefficient above 12.5% averaged about 0.8 fewer puppies than litters below 6.25%, and inbreeding also reduced individual survival. A DNA-based study of 93 female Golden Retrievers found the same direction using measured genomic inbreeding rather than pedigrees. At breed level, a 227-breed analysis found breeds with genotype-based inbreeding above 0.25 had about 29% more veterinary care events than the least inbred breeds.

Is linebreeding the same as inbreeding?

Genetically, yes, they are the same process at different intensities, and the coefficient does not care which word you use. Linebreeding is the term breeders use for mild, deliberate concentration on a particular ancestor, but there is no threshold where one becomes the other. The honest way to describe a mating is its coefficient and the generation depth behind it, not the label.

Why is a pedigree COI usually too low?

Because the maths assumes the founders at the top of the tree were unrelated to each other. In a closed breed built from a small founder base, they usually were related, so every relationship that predates the studbook is invisible to the calculation. Measured DNA typically comes back higher. In the Bullmastiff, where both kinds of estimate exist, the molecular figures were tighter than the pedigree ones.

Sources & further reading

The registry claims on this page come from reading the governing documents themselves: the SABBS Constitution, Bylaws and Code of Conduct in their editions accepted 26 June 2026, every archived SABBS edition back to 2015, the predecessor SABT documents, and the AKC registration rulebook. The health and population-genetics claims come from peer-reviewed research. Where a body publishes no rule, we say so rather than inferring one. Numbers in the text link here.

  1. COI FAQs: Understanding the Coefficient of Inbreeding (generation depth, and the <5% / 5-10% / >10% effect bands), The Institute of Canine Biology. https://www.instituteofcaninebiology.org/blog/coi-faqs-understanding-the-coefficient-of-inbreeding
  2. Pedigree-Based Genetic Diversity in the South African Boerboel Dog Breed (Mabunda, Nephawe, Mtileni & Makgahlela, 2024, Animals 14(6):975): 87,755 studbook records, 1971-2019, MDPI / PubMed Central. https://pmc.ncbi.nlm.nih.gov/articles/PMC10967628/
  3. Inbreeding impact on litter size and survival in selected canine breeds (Leroy, Phocas, Hedan, Verrier & Rognon, 2015, The Veterinary Journal 203(1):74-78), Elsevier / The Veterinary Journal. https://www.sciencedirect.com/science/article/abs/pii/S1090023314004559
  4. Inbreeding depression causes reduced fecundity in Golden Retrievers (Chu et al., 2019, Mammalian Genome 30(5-6):166-172): measured DNA, not pedigree, Springer / Mammalian Genome. https://link.springer.com/article/10.1007/s00335-019-09805-4
  5. The effect of inbreeding, body size and morphology on health in dog breeds (Bannasch et al., 2021, Canine Medicine and Genetics 8:12): 227 breeds, genotype-based inbreeding vs insurance morbidity, Springer / Canine Medicine and Genetics. https://link.springer.com/article/10.1186/s40575-021-00111-4
  6. Registering your puppies with the Royal Kennel Club: the refusal of litters bred from father-daughter, mother-son or brother-sister matings, The Royal Kennel Club (UK). https://www.royalkennelclub.com/dog-breeding/after-whelping/registering-your-puppies-with-the-royal-kennel-club/registering-your-puppies-with-the-royal-kennel-club/
  7. Rules Applying to Registration and Discipline (amended to 1 June 2018): the full registration rulebook, which contains no inbreeding provision, American Kennel Club. https://images.akc.org/pdf/rulebooks/RREGS4.pdf
  8. Bylaws to the Constitution of the South African Boerboel Breeders' Society, accepted 26 June 2026 (Bylaw E, Matings; Bylaw G2, Breeders' Code of Conduct), South African Boerboel Breeders' Society (SABBS). https://www.sabbs.co.za/
  9. Constitution of the South African Boerboel Breeders' Society, accepted 26 June 2026 (Clause 7.2, Misconduct), South African Boerboel Breeders' Society (SABBS). https://www.sabbs.co.za/
  10. Logix for Dogs breeder guide (10 March 2015): the SA Stud Book inbreeding calculator, which displays a planned mating in red above 6%, South African Boerboel Breeders' Society / SA Stud Book. https://www.sabbs.co.za/
  11. Canine Genetic Diversity and Internal Relatedness: why pedigree estimates and measured DNA disagree, UC Davis Veterinary Genetics Laboratory. https://vgl.ucdavis.edu/test/canine-genetic-diversity
  12. Comparative Analysis of Genome Diversity in Bullmastiff Dogs (Mortlock et al., 2016): pedigree and molecular estimates side by side in one breed, PLOS ONE. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0147941
  13. Trends in genetic diversity for all Kennel Club registered pedigree dog breeds (Lewis, Abhayaratne & Blott, 2015): effective population size across 116 breeds, Canine Genetics and Epidemiology / PubMed Central. https://pmc.ncbi.nlm.nih.gov/articles/PMC4579366/
  14. Effective population sizes: the sustainable (>100) and high-risk (<50) lines, The Royal Kennel Club (UK). https://www.royalkennelclub.com/health-and-dog-care/what-we-do-for-dog-health/supporting-scientific-research/kennel-club-health-research/effective-population-sizes/
  15. Boerboel Genetics & Ancestry: the breed-level founder bottleneck, mean inbreeding and effective population size (companion page), Peterbuilt Boerboel. https://www.peterbuiltboerboel.com/the-boerboel-breed/genetics
  16. Boerboel Health & Lifespan: what health testing should actually cover (companion page), Peterbuilt Boerboel. https://www.peterbuiltboerboel.com/the-boerboel-breed/health

Last reviewed September 2026. SABBS amends its Constitution and Bylaws annually, so the registry claims here are tied to the 26 June 2026 editions and should be re-checked against the current documents after each amendment. Spot a genuine error or have a primary source we missed? Tell us via our contact page. We’d rather be corrected than wrong.

Keep reading the breed library

The breed-level picture behind these numbers is on the genetics page, and what “health tested” should actually mean is covered in the health deep-dive.

Read: Boerboel Genetics & AncestryRead: Health & Lifespan

or view available puppies