“Don’t let a big puppy grow too fast or you’ll wreck his joints.” That advice is common enough to feel settled. Part of it holds up extremely well. Part of it is contradicted by the field’s own standard reference work, and that part gets repeated at least as often.
Sorting the two is worth doing, because they tell you to do different things.
What is actually established
The strongest evidence in canine orthopaedics comes from one Labrador colony followed for life.
Kealy and colleagues (1992) took 48 eight-week-old Labradors. One group ate all it wanted. The other got 25% less of the same food until two years old. Hips were X-rayed at 30, 42, 54, 78 and 104 weeks. Later papers on the same colony confirm the dogs were paired with littermates by sex and weight before the groups were drawn. So it is close to a controlled feeding experiment, with genetics balanced across the groups.
Two terms for the table. Hip dysplasia is a hip whose ball and socket do not fit properly. Osteoarthritis, or OA, is the joint wear that follows.
| Outcome | Restricted | Free-fed |
|---|---|---|
| Hip dysplasia at 2 years, OFA method | 7 / 24 | 16 / 24 |
| Hip dysplasia at 2 years, Swedish method | 5 / 24 | 18 / 24 |
| Radiographic hip OA at 8 years | 3 / 21 | 15 / 22 |
| Lifetime hip OA prevalence | 50% | 83% |
| Median age at first radiographic hip OA | 12 years | 6 years |
Those last three rows come from Kealy et al. (2000) and Smith et al. (2006). A six-year gap in the median age at which arthritis first shows on X-ray is a very large effect for a diet change.
One thing to be precise about. The restricted dogs were fed 25% less than a group eating ad libitum, as much as it liked. They were not fed 25% less than a label recommendation. The control group was, in effect, free-fed into an overweight condition. So the finding is that avoiding overfeeding protects joints. It is not evidence that cutting a correctly fed, lean puppy’s food further does anything at all.
Growth rate, or body mass?
This is the honest gap at the centre of the topic, and nobody seems to say it out loud.
The restricted dogs in that colony grew more slowly and stayed leaner for life. Those two move together by design, and we could not find a study that pulls them apart. So when someone tells you that growing too fast damages joints, the evidence behind that cannot separate it from a plainer claim: carrying more weight on young joints damages them.
We think the second is the more defensible reading, and the more useful one. You can see body condition; you mostly cannot see growth rate. But we are giving it as our reading, not as a finding.
Hip dysplasia is inherited first
Any account of this that leaves out genetics is misleading. Here are three heritability estimates from three data sources. Heritability is the share of variation within a group that genes explain.
| Source | Population | Hip | Elbow |
|---|---|---|---|
| Lewis 2013 | 142,287 UK BVA/KC hip scores, 15 breeds | 0.38 mean (0.28–0.48) | 0.218 mean |
| Zhang 2009 | 2,716 research colony dogs, direct laxity measures | 0.54–0.76 | — |
| Hou 2013 | 895,864 OFA records, 74 breeds | 0.22–0.23 | 0.16–0.17 |
These are not in conflict. Heritability belongs to a population, a trait definition and a way of measuring. Sliding-scale measures of joint looseness in a research colony always give higher figures than ranked registry grades in a population already bred for better hips.
One misreading to avoid. A heritability of 0.38 does not mean 38% of your dog’s hip score is genetic. It describes the spread within a group, not one animal.
Laxity is necessary, but not sufficient
The bridge between genetics and feeding is joint laxity, how loose the hip sits in its socket. Vets score it as a distraction index, measured on an X-ray with the joint gently drawn apart. Smith and colleagues (2001) examined 15,742 dogs across four breeds. Every 0.1 rise in that index raised the odds of hip degenerative joint disease by 1.9 to 2.7 times, depending on breed, in dogs aged 24 months or older. Weight and distraction index were both real risk factors in all four breeds.
That gives the defensible framing: the weakness is inherited, and body mass changes how badly and how early it shows. That sentence is supported. Anything stronger is not.
The osteochondrosis claim does not survive
Here the popular story and the specialist literature part company.
Osteochondrosis is cartilage that fails to turn into bone properly. Most of a puppy’s skeleton starts as cartilage and is swapped for bone as it grows. Where that swap goes wrong, a flap of cartilage can lift inside a joint.
Ytrehus, Carlson and Ekman (2007), the standard pathology review, define the first lesion as “a focal ischemic necrosis of growth cartilage initiated by necrosis of cartilage canal blood vessels.” Put plainly: a patch of growth cartilage dies because the tiny vessels feeding it fail. That is a blood supply failure, not a mechanical overload. On the growth question they are blunt: “The majority of the published literature in all species fails to support a direct role of increased growth rate/overnutrition in the etiology of osteochondrosis.”
They also revisit the 1974 Great Dane overfeeding experiment usually cited as proof. Twenty-four dogs were fed either freely or in restricted amounts, on a diet “rich in protein, energy, calcium, and phosphorus.” Energy and minerals moved together, so they cannot be told apart. Osteochondrosis in the joint surface turned up in only five dogs, split almost evenly between the groups. As the reviewers put it, that “fails to support the hypothesis that rapid growth increases the incidence of osteochondrosis in dogs.”
The 2015 update confirms the blood vessel mechanism in pigs and horses. For dogs it says “further studies are needed to reach a conclusion.” Even the epidemiology (the study of how often a disease turns up, and in which animals) hedges. Engdahl and colleagues (2024) worked from over 600,000 insured Swedish dogs. They write that rapid weight gain “has been suggested” to raise risk. They found 3.77 cases per 10,000 dog-years, with males at higher risk, and every breed at raised risk was large or giant. So breed size matters. The growth-rate mechanism is the part with nothing under it.
Elbow dysplasia is a different problem
It gets bundled in, and it probably should not be. Michelsen’s review (2013) notes that elbow dysplasia was first put down to osteochondrosis. Newer evidence points strongly instead to joint incongruity, the bones of the joint not matching each other. One forearm bone outgrows the other (radioulnar length discrepancy), or the curve of the upper arm bone misses the notch it sits in (humeroulnar curvature mismatch). The review makes no claim that growth rate or diet causes it, and we found none elsewhere.
Exercise: what one cohort actually measured
Nearly all the exercise advice you will meet traces back to one prospective study. It recruited puppies, then followed them forwards. Krontveit and colleagues (2012) followed 501 dogs from 103 litters across four large breeds, X-raying at 12 or 18 months. Hip dysplasia overall ran at 24.6%, from 9.5% in Irish Wolfhounds to 36.0% in Newfoundlands.
The findings, all limited to what happened between birth and three months:
- Daily stair use raised risk: odds ratio 1.96 (95% CI 1.14–3.35).
- Daily off-lead exercise on park-like ground lowered risk: OR 0.31 (95% CI 0.15–0.65).
- Being born on a farm rather than in a town home lowered risk: OR 0.34.
The authors’ own conclusion: puppies three months or younger “should not be allowed access to stairs, but should be allowed outdoor exercise on soft ground in moderately rough terrain.”
Two things get turned into something they are not. No exercise link was found for any exposure after three months of age. And the protective factor was free exercise, not restriction. So “keep the puppy quiet” is not what this study says. Birth season also appears in the abstract as protective, but those model terms were marginal, with confidence intervals crossing 1, so we are not presenting it as a risk factor.
Growth plates, and the advice built on them
The growth plates are soft zones near the ends of a young bone where new length is added. They close earlier than the popular advice implies. Von Pfeil and DeCamp (2009) compile the X-ray ranges: lower humerus 5–8 months, lower radius 6–11, lower femur 6–11, tibia 5–12. The tibial tuberosity comes later, 10–12 months, and “potentially 15 to 18 months in giant-breed dogs.” So in a 25–30 kg dog, most limb plates are closed between roughly 5 and 12 months, not 18.
That fits our plate-by-plate closure timetable, which gives 10–13 months for the last plates to close in an average dog, and 18–20 months in giants. The bulk of closure happens well before the final plates finish. It is those final plates the popular advice is quoting.
Two caveats. That is a continuing-education review pulling together older X-ray data, and we found no modern study systematically comparing closure age across size groups. The one recent large-breed X-ray study, Roccaro et al. (2021), looked at when centres of bone formation first appear, between 6 and 16 weeks. Several appeared later than the classic literature states. The textbook numbers are old and do not reproduce perfectly.
Which brings us to the “five minutes of exercise per month of age” rule. Correcting an earlier version of this page: the rule is not internet folklore. The Royal Kennel Club publishes it, as do Blue Cross and the AKC, none of them citing a source. The PDSA states outright that there is no scientific evidence behind it. It may still be sensible caution. It is not evidence, and the research above measured what kind of exercise, never how many minutes. We take the whole rule apart in how much exercise does a puppy need?
Prevalence figures are biased, and by a known amount
Look up hip dysplasia rates for a breed and you usually find registry numbers. They understate the problem.
Paster and colleagues (2005) put a figure on it. Of 93 dogs X-rayed for OFA certification, only 49 had the films actually sent in, and normal X-rays were 8.2 times as likely to be submitted as abnormal ones. In dogs that looked healthy and were examined directly, hip dysplasia ran at 53–73% in Golden Retrievers and 41–69% in Rottweilers, depending on the scoring rules. The OFA registry’s own analysts agree that the public data “is biased … by the voluntary nature of the reporting.”
National screening schemes do better, but they are not clean. Swiss data across 11,136 dogs shows real improvement: German Shepherd Dogs from 46.2% to 18.0% between 1995–99 and 2010–16, Labradors from 16.5% to 2.9%. Its authors still caution that with only 11–31% of dogs X-rayed, the figures do not establish true prevalence.
What we cannot tell you
- Whether growth rate matters on its own, apart from body mass. No study separates them.
- What causes metaphyseal osteopathy, also called hypertrophic osteodystrophy: painful swelling of the bone just below the growth plate in young large-breed dogs. The cause is unclear, and several factors are suspected to act together. The supplement and vitamin C explanations are not established.
- What causes panosteitis: bone pain that shifts from limb to limb. It affects young large-breed dogs from about 6 to 16 months and clears up on its own at skeletal maturity. But per the Merck Veterinary Manual, the suspected genetic, infectious, nutritional and autoimmune causes stay suspected.
What to actually do
- Keep the puppy lean. This is the one step with real numbers behind it. See how much to feed a growing puppy and how to tell if your puppy is overweight.
- Avoid stairs in the first three months, and let the puppy run freely on soft, uneven ground rather than shutting it in.
- Ask the breeder for hip and elbow scores on both parents. The weakness is inherited, and this is the largest single lever. It is pulled before you bring the dog home.
- Treat registry prevalence figures as floors, not estimates.
- Do not assume large-breed puppy food is joint insurance. It governs calcium; portion size governs body mass. Our article on large-breed puppy food covers what the label does and does not promise.