“How big will my Goldendoodle get” is a harder question than “how big will my Labrador get”, and not for the reason people assume. It is not that doodles are mysterious. A Labrador comes with a century of standard record keeping attached to its name. A Goldendoodle comes with a size label invented by a breeder club that openly calls its own numbers rough.

The honest headline is the spread: a Goldendoodle can finish anywhere from about 15 lb to over 90 lb. That span is the finding, not a dodge. Here is what the categories actually say, who decided them, and what the genetics does and does not let anyone promise.

The four categories, and who wrote them

The Goldendoodle Association of North America (GANA) publishes the size varieties nearly everyone quotes:

VarietyHeight at withersWeight
PetiteBelow 14 intypically 25 lb or less
MiniatureOver 14 in, under 17 intypically 26–35 lb
MediumOver 17 in, under 21 intypically 36–50 lb
StandardOver 21 intypically 51 lb or more

Three things about that table matter more than the numbers in it.

Height defines the category, not weight. GANA’s breed standard says so outright: “Height will be the determining factor for accurate size since weights can fluctuate.” Every weight above is prefixed “typically” in the original. Anyone handing you “26–35 lb” as the spec for a mini is quoting a guideline as though it were a rule.

The categories have gaps. Below 14. Over 14 to under 17. Over 17 to under 21. Over 21. A dog measuring exactly 14, 17 or 21 inches lands in no category at all, and Petite has no lower bound. GANA’s two pages even disagree on what the smallest variety is called: “Petite” on one, “Petite Mini” on the other.

GANA gives away the central point. From the same page: “Please be aware that the estimated size of a puppy by a breeder is an educated guess.” And: “a puppy’s final size may vary somewhat from a breeder’s initial calculation.” The body that wrote the categories is telling you not to treat a breeder’s number as a promise.

There is no breed standard

Goldendoodles are not on the AKC’s list of recognised breeds, which covers 200 of the roughly 340 breeds worldwide. The way in for a doodle is Canine Partners, the programme that “welcomes all mixed breed dogs” and lets them compete in agility, rally and obedience. That buys a dog a place to compete, not a breed standard. Goldendoodles are missing from the FCI list too.

GANA is open that this is on purpose rather than an oversight. The Goldendoodle “is not a conformation show breed”, not a breed judged on its looks in a show ring. Dogs “may be bred from the founding breeds or from multigenerational stock,” and “the stud book will remain open.”

Nobody agrees with anybody

If these categories tracked something in the dogs themselves, you would expect different crosses to land on the same scheme. They do not.

The Australian Labradoodle Association of America defines three Labradoodle sizes rather than four (Miniature, Medium, Standard, no Petite) and sets them by height alone. It gives one weight range, 15–65 lb, for the whole breed instead of one per variety. Its standard also says: “Inter-size breeding is acceptable and expected at the moment.” The Australian association, by contrast, does publish weights per size: 7–13 kg, 13–20 kg, 21–30 kg. Two bodies, one cross, opposite views on whether weight belongs in a standard at all.

It goes further down the chain. The Poodle parent is what supposedly sets the variety, and the AKC Poodle standard gives height only, with no weights at all: Standard over 15 inches, Miniature 15 inches or under and over 10, Toy 10 inches or under. Those are cut-offs for the show ring, not lines drawn in biology. And the FCI recognises a fourth Poodle size the AKC does not: a Medium at over 35 up to 45 cm. So a US breeder advertising a “medium Poodle” parent is borrowing a European convention.

Every layer of this naming system is a different committee’s preference.

The generation labels

F1 and F2 are real genetics terms. F1 is the first cross: here, a Golden Retriever bred to a Poodle. F2 is two of those F1 doodles bred together. F1b is not a real term. It does not appear in the NHGRI genetics glossary, which defines around 250 terms. Standard genetics calls it a backcross: a doodle bred back to one of the parent breeds. The “b” suffix, and spin-offs like F1bb and F2b, are breeder-community coinages.

That matters because of what the coinage gets used to imply. GANA’s own definition reads: “An F1b is a Goldendoodle backcrossed to either a Golden Retriever or a Poodle.” Popular usage assumes the Poodle. But under the official wording an F1b could be three-quarters Golden Retriever, which would trend larger. So “F1b are smaller” is not merely unsupported. It contradicts the definition it depends on.

We went looking for evidence that generation changes how predictable size is. We found none: no study, no registry dataset, no published analysis. GANA’s own generations page makes no size claim at all. We would rather tell you that than hedge.

What the genetics actually supports

This part is real science, and it is genuinely impressive.

Dog body size runs on a surprisingly small number of genes, each with a large effect. Boyko and colleagues (2010) found that for most traits of body shape, “a small number of quantitative trait loci (≤3) explain the majority of phenotypic variation.” A locus is simply a spot on a chromosome. For size, one marker at the IGF1 gene reached R² = 50%. It accounted for half the variation on its own. Rimbault and colleagues (2013) reported that seven markers near six genes explain “46–52.5% of the variance in body size of dog breeds.” Plassais and colleagues (2019) put it at “greater than 90% of body size variation in dogs,” with variants in just 14 genes covering as much as 95% in purebreds.

For contrast, Hayward and colleagues (2016) note that human height “is associated with nearly 700 variants, which cumulatively explain only ∼20% of the variation in adult stature.” Dog size is a far simpler genetic puzzle than human height.

And here is the finding that actually answers the question. Those high percentages come from purebred dogs, where the breed name itself already carries much of the answer. Run the same class of model on dogs outside the breeds and it falls apart. Hayward reports that in village dogs “the linear model only explains 30–40% of body size variation,” and Boyko’s correlation drops from 0.85 in breed dogs to 0.50 in village dogs.

Village dogs are free-breeding mixes, not F1 doodles, so this stands in for the doodle case rather than measuring it. But the direction is not in doubt, and it is the paper’s own result, not our reading of it: predicting size gets substantially harder once you leave the standard breeds.

What we can reason but not claim. Size rests on few genes of large effect, so an F1 cross carries one Golden copy and one Poodle copy at nearly all of them: heterozygous, in the jargon. If those copies simply add together, F1 littermates should cluster fairly tightly around the midpoint between the parents. It should be the later generations (F2 and backcrosses, where the copies get shuffled and dealt again) that throw the wider litters. We are labelling that as inference from the genetics rather than a finding. Nobody has published adult size distributions for deliberate crossbreeds. And we could find nothing describing how the small-size IGF1 variant behaves when paired with a large-size one, which is the very thing the midpoint argument hangs on.

So: do not let anyone tell you F1 doodle sizes come out in two clumps, one big and one small. Nothing supports that. And do not let anyone tell you the science proves they land on the midpoint either.

What large datasets show

Doodles do appear in veterinary records, though only lumped in with other crosses. The VetCompass UK demography study recorded designer crossbreds at a median adult bodyweight of 13.40 kg. The middle half of those dogs fell between 10.22 and 19.00 kg. That band is the interquartile range, or IQR. The full range ran from 2.15 to 57.00 kg. That range pools every designer cross and is pulled about by the extremes, so the IQR is the more responsible number, still nearly a two-fold span. The Cockapoo was the third most common type among dogs under one year old, at 6.2%. Nationwide’s insured-dog analysis found Goldendoodles up 347% in relative popularity across 2013–2021 while Standard Poodles fell 32%. That is insurer data, not a head count of every dog.

There is a built-in reason doodle growth charts are weaker than they look. The best-validated veterinary growth standards, Salt and colleagues (2017), state that “mixed breed dogs, and those from all other breeds, were excluded”. The method also works by first looking up a dog’s expected adult size from its breed average. For a Goldendoodle there is no breed average to look up.

What we would actually do

  • Ask for the parents’ adult weights, not the puppy’s category. Two measured numbers beat a label that has gaps in it.
  • Weigh your puppy and follow its own curve. Growth timing tracks adult size rather than ancestry, so a weight taken at a known age tells you something the label cannot. Our chart of adult-weight share by size class gives the fractions that method needs. Our puppy size calculator runs it for you, and for designer crosses it reports low confidence and a wider range, deliberately.
  • Treat the variety name as a rough band. Our own dataset lists a generic Goldendoodle at 25–90 lb precisely because the label does not narrow it.
  • Ignore paw size. We looked for evidence and found none either way. That is an absence, not a debunking. See do big paws mean a big dog?.

For the broader method on crosses, see how to predict a mixed-breed puppy’s size, and our methodology for where these estimates are weakest.