Most answers to this question are a table on the back of a bag. That table is a fine place to start and a bad place to stop. The reason is not that manufacturers are careless. It is that there is no single right answer for a given weight of dog.
Here is the sum you can actually do, what happened when someone checked it against real puppies, and what to steer by instead.
Start with the bag, then stop trusting it
Feeding directions are not standardised. The 2021 AAHA Nutrition and Weight Management Guidelines put it plainly: owners should ideally work the amount out themselves rather than lean on the pack, “because AAFCO does not standardize this process.”
How much do bags vary? The clearest figure we found is for weight-loss diets rather than growth diets. Linder and Freeman (2010) looked at 93 commercial weight-management products. For the same standard 36.4 kg dog, the amounts they told you to feed ran from 0.73 to 1.47 times resting energy requirement, a two-fold spread across products, for one identical animal. Nobody has run the same check on puppy food as far as we can see, so we will not claim the number carries over. But it is the right amount of doubt to bring to a bag.
The equation that actually exists
The National Research Council, or NRC, publishes an energy requirement equation for growing dogs. We could not get hold of the NRC volume itself, so we are citing two peer-reviewed papers that print it word for word, Böswald et al. (2019) and Bradley et al. (2021):
ME (kcal/day) = 130 × BW^0.75 × 3.2 × (e^(−0.87p) − 0.1)
p is the puppy’s weight now divided by the adult weight you expect it to reach. BW is the puppy’s current weight in kilograms, not the adult weight. That second point is the part almost every other version of this sum gets wrong.
Here is what the multiplier does as the puppy grows:
| Fraction of adult weight (p) | Multiplier on 130 × BW^0.75 | Effective kcal coefficient |
|---|---|---|
| 10% | 2.61 | 340 |
| 25% | 2.25 | 293 |
| 37% | 2.00 | 260 |
| 50% | 1.75 | 228 |
| 75% | 1.35 | 175 |
| 100% | 1.02 | 133 |
Two things are worth noticing. The multiplier settles at about 1.0 once the dog is grown, which means the 130 is simply adult upkeep. WSAVA gives 130 × BW^0.75 for an active adult dog and 95 for an inactive one, so the equation hangs together. And the familiar rule of thumb, “twice maintenance early, tapering to once”, only lands exactly at around 37% of adult weight. Before that it is more than double. After that, less. We went looking for where that rule came from and found nothing, so treat it as a summary of this curve rather than a source in its own right.
A worked example
A Labrador puppy weighs 10 kg and should finish at 30 kg. So p = 0.33.
- 10^0.75 = 5.62
- 130 × 5.62 = 731
- multiplier at p = 0.33 is 3.2 × (e^(−0.29) − 0.1) = 2.07
- 731 × 2.07 ≈ 1,510 kcal/day
If the food runs 380 calories per cup, that is about four cups a day, split across three meals. Now hold that number loosely, because of what follows.
A second equation, from a different continent
FEDIAF, the European pet food trade body, publishes its own growth equation in its 2025 guidelines. It is worth setting beside the NRC one, because it is not a variation on the same formula. It is a different shape of curve arriving at the same kind of answer:
ME (kcal/day) = [254.1 − 135.0 × p] × BW^0.75
Same two inputs, same power of 0.75. But the NRC equation curves down as the puppy nears adult size, steeply at first and then more gently. FEDIAF’s drops in a straight line. Two separate bodies looked at one problem and did not even agree on the shape.
Set the two side by side:
| Fraction of adult weight (p) | NRC | FEDIAF | FEDIAF vs NRC |
|---|---|---|---|
| 10% | 340 | 241 | −29% |
| 25% | 293 | 220 | −25% |
| 37% | 260 | 204 | −21% |
| 50% | 228 | 187 | −18% |
| 75% | 175 | 153 | −13% |
| 100% | 133 | 119 | −10% |
FEDIAF sits below NRC at every stage of growth, and the gap is widest when the puppy is youngest. Hold that thought. It matters for the next section.
Three schedules, and they miss in the same direction
The Merck Veterinary Manual gives a third rule, and a much simpler one. Start from resting energy requirement (what a dog burns doing nothing at all), which is 70 × BW^0.75. Multiply by 3 for a puppy under 4 months, and by 2 after that. So 210 × BW^0.75 under four months, and 140 over.
Now run all three on the same animal. That 10 kg Labrador puppy from above, headed for 30 kg:
| Method | Coefficient | Estimate |
|---|---|---|
| NRC growth equation | 270 | 1,517 kcal/day |
| Merck, under 4 months | 210 | 1,181 kcal/day |
| FEDIAF growth equation | 209 | 1,176 kcal/day |
| Merck, over 4 months | 140 | 787 kcal/day |
That tells you more than a flat “the sources disagree”, and it is why the second equation was worth adding. They do not disagree evenly. FEDIAF and Merck’s under-four-months figure land within five calories of each other for this puppy: two sources, two continents, two methods, in near-perfect agreement. NRC is the odd one out, and it is the high one, by about 22% for this dog.
That still does not make NRC wrong. It is the most carefully derived of the three, and the only one written as a continuous function of growth stage. It moves smoothly instead of jumping on a birthday. But when you hold three numbers and two of them agree, it helps to know which one stands apart, and which way.
The best available test says the equation overfeeds
This is the finding that should change how you use the number above.
Bradley and colleagues (2021) followed 17 Norfolk terrier puppies from 10 to 52 weeks. Each puppy was fed whatever amount held it in ideal body condition. The team then set that real intake against what the NRC growth equation predicts. The puppies ate less than the equation called for at every point measured, and the gap was widest at 10 weeks. Feed by the equation, the authors conclude, and dogs “would have been in positive energy balance, possibly leading to obesity”, taking in more than they burn.
The limit is real and worth saying out loud: one small breed, seventeen dogs. That is not proof the equation overfeeds every puppy. It is the only direct test we could find, and it points one way.
But notice what happens when you put it next to the section above. Three separate things line up:
- FEDIAF’s equation sits below NRC’s at every stage, and by the widest margin in early growth.
- Merck’s under-four-months figure lands almost exactly on FEDIAF’s, leaving NRC alone on the high side.
- The one real-world test of NRC found it called for more than puppies ate at every point measured, and again, the gap was widest at the youngest age.
Three different kinds of evidence (a rival guideline, a clinical convention and a feeding trial) agree on which way the error runs and on its shape. That is a stronger position than any one of them alone. It is why we would treat the NRC number as a ceiling rather than a target.
We would still not tell you to feed a set percentage less. Seventeen Norfolk terriers cannot fix a correction factor for every puppy. When three sources suggest a number runs high, the honest response is not a different number. It is to watch the dog.
Two dogs the same weight can need very different amounts
The WSAVA Nutritional Assessment Guidelines say a label figure or a formula “may be used as a starting point for energy allowance since energy requirements can vary by … 30% in either direction for dogs.”
What people have measured is wider still. Bermingham and colleagues (2014) ran a meta-analysis, pooling the results of many separate studies: 70 treatment groups covering 713 dogs. Mean maintenance energy requirement, meaning resting needs plus everyday moving about, came to 142.8 ± 55.3 calories per kg of bodyweight to the power 0.75 per day. The full range ran from 54.5 to 441.1. Neutered dogs sat lower than intact ones, 146.4 against 195.7.
Two caveats, which we would rather state than bury. That dataset takes in colony, kennel and working dogs, sled dogs among them. So the eight-fold total range is not the gap between two pet dogs on your street. And in adult dogs the NRC equations came out low, not high. The error runs one way in puppies and the other way in adults, and the two findings should not be blended.
So steer by condition, not by the bowl
Every body we read lands in the same place. The sum is an opening bid. The dog is the measuring device.
WSAVA sets the target at a body condition score of 4 to 5 out of 9, and says you get there by hand: palpation (feeling the dog) “is required”, not looking at it. AAHA (2021) wants owners taught to score condition so they can “adjust intake as needed,” and for giant-breed puppies it asks for “maintain[ing] a lean BCS.” Merck says weigh weekly through growth and adjust so the puppy gains a little each week. And the Bradley study above is exactly this method run as a trial. Puppies got whatever amount held them in good shape, and that amount kept coming out below the formula.
There is an important limit to this. Body condition scoring was validated in adult dogs. Using it on a growing puppy stretches the method past where anyone has tested it. What to do about that is in how to tell if your puppy is overweight, which is the other half of this article.
How often to feed
Standard veterinary guidance, from Merck’s feeding practices chapter: three times a day from weaning to 6 months, twice a day from 6 to 12 months, and small-breed puppies may need more than three feedings.
We looked for a controlled trial of meal timing in growing dogs and found none. This is clinical convention: sensible, near-universal, and not the same thing as evidence.
Free feeding, meaning food left down all day, is a different question, and there the evidence is real. Merck notes that most adult dogs cannot be fed free choice “without becoming obese,” and the study in the next section was, in effect, a free-feeding trial.
What overfeeding actually costs
Kealy and colleagues (1992) took 48 Labradors at eight weeks old. One group was fed ad libitum, as much as it liked. The other got 25% less of the same food, until two years old. Both groups had their hips radiographed (X-rayed) at 30, 42, 54, 78 and 104 weeks. Hip dysplasia turned up in 7 of 24 restricted dogs against 16 of 24 free-fed dogs by the OFA method, and 5 of 24 against 18 of 24 by the Swedish method.
The same colony was then followed for life. Median lifespan came out at 13.0 years in the restricted group against 11.2 in the controls, and osteoarthritis, wear-and-tear joint disease, set in later. For anyone worried about feeding less, one line matters. Lawler and colleagues (2008) report that long-term restriction “did not negatively affect skeletal maturation, structure or metabolism.”
One correction, because this study gets misquoted constantly. The restricted dogs were fed 25% less than a free-fed control, not 25% less than a recommended amount. Those controls ate all they wanted until they were more than three years old, and across ages 6 to 12 they averaged a body condition score of 6.7 out of 9, which is overweight. So the finding is that avoiding overfeeding adds years and protects joints. It is not a finding that cutting a well-fed dog’s food does any good. If you already feed to a lean body condition, this study is not telling you to cut further. We go through what that Labrador colony showed about joints over a lifetime, including the caveat that makes it weaker evidence than it is usually presented as.
Three things that are not true
“Feeding less will stunt him.” Merck states that it is hard to stunt a puppy fed a complete and balanced growth food, and the Labrador colony above found no harm to skeletal maturation (how the bones develop and finish) from a lifetime at 25% less.
“A raised bowl is better for a big puppy.” The evidence runs the other way. Glickman and colleagues (2000) followed 1,637 large and giant breed dogs. A raised bowl came out as one of the risk factors for gastric dilatation-volvulus (bloat, where the stomach fills with gas and twists). They put about 20% of cases in large breeds, and 52% in giant breeds, down to it.
“Once a day is healthier.” There is a Dog Aging Project paper linking once-daily feeding to better health markers. But it is cross-sectional (a snapshot in time) and owner-reported. Its own authors call for longitudinal data, meaning dogs tracked over years, before anyone treats this as cause and effect. And it looked at adult pets, not growing puppies. It does not make once a day right for a puppy, and it cuts against standard advice for the young.
The short practical version
- Run the equation, or use the bag, to get a starting number.
- Split it across three meals to 6 months, two after that.
- Weigh weekly and check condition by hand.
- Adjust the amount, usually down from the formula, until the ribs are easy to feel and there is a waist and a tuck.
- Re-run the number every few weeks, because the puppy’s weight is the input and it keeps changing.
If you do not know your puppy’s expected adult weight, which the equation needs, our puppy size calculator will estimate it across our whole breed dataset, with the error bars published rather than hidden.
Once the growing is finished, the sum changes and the sources stop agreeing with each other. Our dog calorie calculator handles the adult version, and shows you the spread rather than picking one figure out of it.