Run this breed through the screening databases and one number jumps out. Nearly nine in ten tested Havanese carry a chondrodysplasia variant. No breed scores higher.
Reported bare, that reads like a catastrophe. It is not. It is the reason the legs are short, which is the reason the dog looks like a Havanese. The database is describing the breed standard in the language of DNA.
So this page has two jobs: explain why the alarming number is not alarming, then point at the number that is. That one is in the mouth.
The Havanese has held 25th place on the AKC’s registration list since 2023.
A height with a disqualification, and no weight at all
Both standards we could open measure this dog and refuse to weigh it. The American one wants “The ideal height is between 9 and 10½ inches, with an acceptable height range from 8½ to 11½ inches”, and disqualifies outside it, except that the floor does not apply under twelve months. The international one asks for 23 to 27 cm with a tolerance of 21 to 29. Neither prints a weight anywhere.
That matters for the panel above. Those weights cannot have come from a standard, and the male-female difference our dataset shows is calculated rather than counted; our methodology page sets out how.
| What the figure describes | Value | Where it comes from |
|---|---|---|
| Ideal show height | 9–10½ in | AKC standard |
| Acceptable show height | 8½–11½ in | AKC standard, with disqualification outside |
| International height | 23–27 cm, tolerance 21–29 cm | FCI Standard No. 250 |
| Show weight | not specified | neither standard gives one |
| Mean height, dogs with two copies of the short-leg variant | 26.4 cm | Norwegian study, dogs with a height recorded |
| Mean height, dogs with one normal copy | 31.3 cm | same study |
One footnote on origin. The American standard calls this “The native dog of Cuba”. The international standard files the breed under Cuba, then contradicts the name in its own history note: roots in the western Mediterranean, and the Havana connection a legend grown out of the dogs’ tobacco colour.
The short legs are a gene, and nearly every dog has it
OFA holds a DNA result called CDPA: chondrodysplasia, meaning cartilage that grows in an altered pattern. Of 133 Havanese tested, 11.3% were clear, 2.3% carried one copy and 86.5% carried two. Add the last two together and 88.7% carry the variant at all. That is first among the breeds OFA ranks on this test. We do not hold how many breeds that is, and a DNA test lists only breeds whose owners submit, so treat the placing as decoration and the split as the finding.
An extra working copy of a growth gene called FGF4 sits on chromosome 18 and tells the long bones of the leg to stop early, while the body stays full size. UC Davis’s genetics laboratory describes CDPA as producing “a short-legged phenotype characteristic of many dog breeds”, inherited as autosomal dominant. One copy is enough.
Norwegian researchers put a number on how fixed it is. Genotyping 355 Havanese at a marker linked to the retrogene, they found frequencies of 0.025 for the normal version and 0.975 for the variant. Twenty-five copies in a thousand. Among the dogs whose owners recorded a height, the few carrying one normal copy were bigger: means of 31.3 cm and 26.4 cm for one copy and two, a gap of 4.9 cm at p below 0.0001. Those dogs were also born on average 4.7 years earlier than the rest, and the normal version has kept getting rarer across two decades.
We cannot reconcile the two samples. The Norwegian frequency implies almost no clear dogs; OFA’s American submissions put 11.3% of 133 in the clear column. They agree on direction, not on level.
The variant that hurts discs is a different one
Dogs carry two of these extra FGF4 copies, on different chromosomes. Chromosome 18 gives CDPA. Chromosome 12 gives CDDY, which the same laboratory says brings “premature disc degeneration and increased susceptibility to disc herniation” along with the short legs: dominant for the disc disease, semi-dominant for height. CDDY is what makes a Dachshund’s spine fragile.
On CDDY the Havanese is close to clean: OFA holds 233 tests, with 6.0% carrying one copy and 0.4% carrying two. Anyone telling you a Havanese has a Dachshund’s back is running the two variants together.
The gap between them has been measured. One study put both copies in the same model of 569 dogs operated on for disc disease. The chromosome 12 copy raised the odds of herniation roughly eighteen-fold: odds ratio 18.42, on a confidence interval of 7.44 to 50.26. The chromosome 18 copy is the one this breed is saturated with. It had “only a minor effect” on how young a dog was at its first disc surgery, and “no effect” on disc calcification. Not quite nothing, then, and the authors say so: other things may modify risk inside a breed, “including the FGF4 retrogene on CFA18”.
The part that is not just breed type
The trait is still not free. The breed club’s own definition is blunt about the mechanism: the growth plates of an affected dog “close prematurely and sometimes unevenly”, which “can cause short straight legs, bowed legs or one straight leg and one bowed leg.” Unevenly is the word doing the work. A leg whose two bones stop growing at different moments does not end up straight.
That collides with the standard, which demands “Forelegs are straight when viewed from any angle” while crediting the springy gait to “the short upper arm”: shortened bones and faultless alignment at once.
How often does the alignment fail? The most-quoted figure is that 44% of Havanese had bowed, shortened or asymmetric forelegs in a 2004 owner survey run by the American breed club. It reaches us second-hand, through a 2007 heritability study and again through the Norwegian paper. Neither says how many owners answered, and the survey is unpublished, a famous number with no denominator we can show you.
That 2007 study did produce a heritability, though not for the legs on their own. It scored four things as one (skeletal, eye, heart and liver abnormalities) and put the heritability of that combined score at 0.36, on standard errors of 0.27 and 0.16 for its two scoring methods. About a third of the difference between dogs traced to genes, across all four at once.
The club’s position is that dogs with clinical chondrodysplasia “should not be used for breeding”. That judges bent legs, not the genotype 88.7% of the breed carries. Ask which a breeder means, and look at the puppy’s front legs from the front and the side.
The teeth are the finding nobody quotes
This is the number that deserves the attention the chondrodysplasia result gets.
28.5% of 621 Havanese failed OFA’s dentition evaluation, fifth among the breeds ranked on that test. For scale, the French Bulldog sits tenth at 17.3% of 404 dogs and the Poodle twenty-eighth at 6.9% of 1,851.
Before that means anything, you need to know what the test looks for. OFA states its purpose plainly: “The purpose of the database is to certify dogs with all adult teeth fully erupted. The database does not certify overall dental health, misaligned teeth, or dentition in accordance with a breed standard.” A vet counts the teeth and marks each one M for missing or P for a baby tooth still in place.
So this is a count, not a diagnosis. A failing dog is missing adult teeth, or has teeth that never erupted, or still holds puppy ones. It says nothing about gums. Owners in the 2004 club survey reported dental deviations in 18% of their dogs, the same direction from a softer source.
An all-dogs comparator comes with a warning. Slovenian clinicians examined 627 dogs referred for dental treatment between January 1999 and March 2000 and found abnormalities in 288 of them, 45.93%. Within that 288, 93 had hypodontia (missing teeth), 99 a bent root, 34 fused roots, 22 extra teeth, 18 extra roots. Watch the switch: 45.93% is out of 627, everything after it out of 288. And 627 dogs sent to a dental clinic are not 627 ordinary dogs, so whether 28.5% is high, that cannot say.
Gum disease is a separate question with a better dataset. Following 22,333 British dogs for a year, the VetCompass team recorded dental disease in 2,797, or 12.5%. Risk ran highest in the Toy Poodle at 3.97 times the crossbred rate, then the King Charles Spaniel at 2.63, the Greyhound at 2.58 and the Cavalier at 2.39; lowest in the German Shepherd Dog at 0.25, the French Bulldog at 0.43, the Staffordshire Bull Terrier at 0.45 and the Labrador at 0.49. Only the four highest and four lowest are named (eighteen breeds in all came out at raised risk), and the Havanese is in neither group. Small mouths score badly there, so assume nothing protective.
Eyes: a clean-looking number that undersells the risk
The registry’s eye figure is one of the best in this dataset: 3.1% of 16,198 evaluations graded abnormal. Now the correction, and it is a big one.
Norwegian records of 2,581 Havanese eye screenings between 2003 and 2022 put cataracts at about 5% of examinations, and explain why any single screening figure reads low. The average age at examination was 3.4 years, standard deviation 1.9. The average age at which cataracts appeared was 4.8 years. Only 7.7% of examinations were done on dogs of seven or over.
Dogs are being checked, on average, more than a year before the disease usually shows up. So a clear certificate on a three-year-old parent is a weaker promise than it looks, and one clear result on a grandparent of eight beats two on young dogs.
Lifetime risk is higher again. The same paper reports that in North American veterinary records the Havanese came out second among all breeds for cataract, at an estimated 11.57%. Owners in the 2004 club survey described atypical lenses in 11% of their dogs, the same neighbourhood from another direction.
The genetics are early. That team’s gene scan set 57 controls against 27 cortical, 23 anterior suture line and 7 posterior polar cases, and only the posterior polar form produced significant markers, on chromosomes 20 and 21. Seven cases is tiny. There is no test to buy.
One more limit. OFA’s eye exam uses “indirect ophthalmoscopy and slit lamp biomicroscopy” and leaves things out on purpose: no tear measurement, no ulcer stain, no eye pressure. A pass means no listed inherited disease was seen that day.
Ears: a rare piece of unambiguous good news
0.2% of 9,249 BAER hearing tests came back abnormal. The breed club has dropped it as a certificate requirement for exactly that reason (the “OFA-BAER test is no longer required for CHIC because its incidence is so low”), though it still recommends one. OFA notes that a dog deaf in one ear “appear[s] to hear normally” unless BAER is run. Half deafness is invisible at home, which is why the test exists at all.
Joints: the quiet section of the file
Toy breeds usually have a bad kneecap story. This one does not. These are OFA’s breed statistics as of 2 August 2026.
| Test | Dogs evaluated | Graded abnormal | For comparison |
|---|---|---|---|
| Hips | 7,623 | 9.3% | 139th of 219 breeds ranked |
| Kneecaps (patella) | 11,342 | 3.4% | Pomeranian 29.1% of 3,481; Yorkshire Terrier 17.0% of 3,169 |
| Elbows | 4,226 | 6.5% | we hold no count of breeds ranked, so no placing |
| Legg-Calvé-Perthes | 3,384 | 0.1% | Yorkshire Terrier 3.0% of 574 |
Legg-Calvé-Perthes is the hip’s ball losing its blood supply in a young small dog; the club’s guide puts the signs between five and twelve months. One dog per thousand tested here, against thirty in the Yorkshire Terrier. The 2004 owner survey said 1% for it, 7% for hips and 5% for kneecaps.
Kneecaps have a proper population comparator. Across 210,824 dogs in English practices, 1.30% had a luxating patella in a single year, on a confidence interval of 1.21 to 1.39. Against crossbred dogs, odds ran 6.5 times higher in Pomeranians, 5.9 in Chihuahuas, 5.5 in Yorkshire Terriers and 3.0 in the Bichon, this breed’s nearest relative there. The Havanese is not among the breeds that paper names. And a one-year figure and a lifetime screening figure answer different questions. Do not stack them.
Two cautions belong on every number above, and they pull opposite ways. Paster and colleagues measured the submission bias rather than assuming it: hips that looked normal were 8.2 times more likely to reach the registry, so 9.3% is a floor. But a dog can be graded dysplastic off an X-ray and walk perfectly well for life. Never turn a radiographic percentage into a share of dogs that limp.
Thyroid: watch the middle column
823 Havanese have thyroid panels on file. 7.3% were abnormal and 7.7% came back equivocal. The uncertain results slightly outnumber the confirmed ones, and that middle column is what makes the row worth reading. The 7.3% is ordinary, sitting just above two breeds nobody calls thyroid cases: the Cane Corso at 7.2% abnormal plus 9.3% equivocal on 237 dogs, the Beagle at 6.7% plus 10.7% on 616.
Equivocal is neither a diagnosis nor an all-clear. OFA defines it as dogs “with laboratory results that are questionable, therefore not definitive”. The disease being hunted is autoimmune thyroiditis, where the immune system attacks the gland that sets the body’s metabolic pace. It usually declares itself between two and five years old, which is why an early test lands in the middle: antibodies may still be rising when the blood is drawn. OFA wants a test yearly to age four, then every other year, and certifies nothing before twelve months. If a breeder shows you one panel from a young dog, ask when it was run and whether it was repeated.
What they are actually like
Start with the ceiling on every breed temperament claim. Surveying 18,385 dogs and genotyping 2,155 of them, a 2022 genomics team found breed accounted for only about 9% of the behavioural difference between one dog and the next. One part in eleven. A weak hint, not a forecast.
This breed does have data, though. The most-cited survey of canine aggression, Duffy and colleagues, ran two samples: 11 breeds through clubs, 33 online. The Havanese is in the online 33, with 73 dogs, averaging 3.05 years old. Its row in the table counting dogs that ever snapped, bit or tried to bite:
| Directed at | Havanese | Rate | Same table, for scale |
|---|---|---|---|
| Strangers | 2 of 73 | 2.7% | Dachshund 14 of 68 (20.6%) |
| The household | 0 of 73 | 0% | Chihuahua 3 of 56 (5.4%) |
| Other dogs | 3 of 73 | 4.1% | Chihuahua 10 of 56 (17.9%) |
| A dog in the same home | 1 of 56 | 1.8% | Dachshund 5 of 57 (8.8%) |
Note the last row’s denominator. Only multi-dog households can answer it, so 56 rather than 73. That switch is how a wrong number gets copied out of a table.
Three limits: owners chose to take part, the threshold is one maximum score on one question, and the paper’s text never discusses this breed at all.
The written character matches. The American standard asks for “friendly, playful, alert and intelligent with a sweet, non-quarrelsome disposition”. The international one is warmer: “Affectionate, of a happy nature, he is amiable, a charmer, playful and even a bit of a clown.” It adds that the breed trains easily as an alarm dog, a polite way of saying it will tell you when the post arrives.
Growing up small
A dog finishing near twelve pounds skips nearly everything written about puppy growth, because almost all of it was written for big dogs. No long skeletal build-out to slow, no calcium ceiling to police. How puppies finish growing covers what a projection can do.
What no small breed skips is adolescence. Asher and colleagues followed 285 dogs (retrievers, German Shepherds and crosses) by questionnaire at 5, 8 and 12 months. Carers rated trainability lower at 8 months than at 5; trainers rated the same dogs higher. A separate group of 93 dogs took a hands-on test at 5 and 8 months only, and that smaller group owns the well-travelled detail: obedience to a “sit” dropped when the dog’s own carer asked and held up for a stranger. No Havanese in either group.
The Generation Pup cohort shows what owners actually complain about, and the shape is a hump. Barking led at every age: 7.9% of 1,718 owners at six months, 13.4% of 1,219 at twelve, easing to 10.5% of 684 at eighteen. The authors report that “barking, pulling on the lead, recall issues and jumping up at people all peaked in 12-month-old dogs and declined thereafter.” For a breed built to announce visitors, the first birthday is when to have a plan rather than a temper.
The hypoallergenic question, tested on this breed by name
Havanese are sold as the safe dog for allergic households. Two studies bear on that, and one has the breed on its list.
A US team sampled house dust in 173 one-dog homes and measured Can f 1, the main dog allergen. Because nobody agrees which breeds count, they tried four definitions. Their table of breeds marketed as hypoallergenic includes the Havanese, marked under three of the four.
None of the four made a difference. Read the numbers with their denominator attached: homes where allergen was detectable at all, split by whether the dog was let into the baby’s bedroom. Micrograms per gram, hypoallergenic against not:
| Scheme | Dog let in the bedroom | Dog kept out |
|---|---|---|
| First | 2.43 vs 3.07 | 1.27 vs 0.64 |
| Second | 1.46 vs 3.07 | 1.00 vs 0.64 |
| Third | 1.46 vs 2.88 | 1.00 vs 0.73 |
| Fourth | 2.58 vs 2.56 | 3.02 vs 0.72 |
Groups are small (5 to 14 hypoallergenic homes per scheme), and none of the eight comparisons reached significance. Look at the right-hand column, the half nobody quotes: where the dog was kept out of the bedroom, the marketed breeds’ homes measured higher every time. Their summary: “No classification scheme showed that the level of dog allergen in homes with hypoallergenic dogs differed from other homes.”
A Dutch group went at coats instead, comparing 196 dogs of four breeds sold as hypoallergenic against 160 controls. The marketed group carried more allergen: 2.26 against 0.77 micrograms per gram in hair, 27.04 against 12.98 in coat. Their verdict was flat: “There is no evidence for the classification of certain dog breeds as being ‘hypoallergenic.’”
Those four were Poodles, Labradoodles, Spanish Waterdogs and Airedales. The Havanese was not among them, and nobody has swabbed this breed’s coat. A long silky coat brushed rather than shed does leave less hair on the sofa. Loose hair and airborne allergen are not the same substance. If someone at home reacts to dogs, spend time with the individual animal first.
Lifespan, and an honest blank
We could not find a published lifespan for this breed, and we looked hard.
The UK life tables, built from 30,563 dead dogs, give life expectancy at birth of 11.23 years across everything, but they break out only 18 breeds, and the Havanese is not one. The largest longevity paper, covering 584,734 dogs, reports 12.5 years overall, 12.7 for pedigree dogs and 12.0 for crossbreeds, then puts its per-breed figures in a supplementary table we could not open. Nowhere in the part we could read does a number for this breed appear.
The closest usable figure describes size, not breed. Clinical records on 13,292,929 dogs give 12.69 years at birth for all dogs, on 12.68 to 12.70, and 13.36 years for the toy category on 13.33 to 13.38. Five weight bands, not breeds, so it is a toy-dog number a Havanese sits inside.
Everywhere else you will read 14 to 16 years. We could not find the measurement.
What a breeder should be able to show you
The American breed club sets three certificate requirements: OFA hips after two years, OFA patellas after one, and an eye exam by a board-certified ophthalmologist, renewed yearly.
Read the club’s own caveat first. “It is not necessary for the Havanese to pass those tests as the intent of the CHIC program is to simply encourage widespread testing and the open disclosure of results.” A certificate proves the tests were done and published, not that they were good. Ask to see them.
Then notice what is missing. Dentition is not on the list, and that is where the breed places fifth. Neither is the chondrodysplasia DNA test. The two OFA results that most distinguish a Havanese sit outside the scheme, so ask for them by name.
UC Davis’s genetics laboratory lists what it offers for the breed, mostly coat and colour work. Its one health test is prcd-PRA, a form of progressive retinal atrophy, where the light-sensing cells at the back of the eye slowly die off. Worth asking about alongside the annual exam, because a look and a DNA test answer different questions.
The club’s condition summaries are the plainest short reference for owners. On liver shunts (a blood vessel routing blood past the liver instead of through it), the club says most affected dogs show signs before six months, and lists them: poor appetite, dullness, weakness, poor balance, disorientation, seizures. A puppy doing any of that is a same-week appointment, not a wait-and-see.
What we could not verify
That the Havanese is the national dog of Cuba. Repeated everywhere. We found no official designation. The AKC standard calls it “The native dog of Cuba”; the FCI treats the Cuban origin as a legend built on coat colour.
The denominator behind the 44% foreleg figure. Two peer-reviewed papers cite it. Neither says how many owners answered, and the 2004 survey is unpublished.
Whether OFA’s 11.3% clear and Norway’s 2.5% normal-allele frequency describe the same thing. They cannot both be typical. We have reported both.
A lifespan for the breed. The per-breed table we needed is inside a supplementary file we could not open.
Whether 28.5% failing dentition would hold in dogs nobody screens. Registry figures come from owners who chose to submit. There is no unscreened comparison.
A month-by-month weight chart. Every one circulating cites nothing. The milestones above come from our model, run on this breed’s adult weights.
“Havanese need X minutes of exercise a day.” No study, here or anywhere.
Allergen measured in Havanese coats. Nobody has sampled it directly.
Eight blanks, and together they make the point. This breed is unusually well screened (sixteen thousand eye exams, eleven thousand kneecaps, nine thousand ears), and the screening still misses most of what a buyer wants to know. The teeth, the front legs and the date on the eye certificate are where your questions belong.