Food & Nutrition

Grain-Free Dog Food: What the Science Actually Says

Updated July 28, 2026

cottonbro studio / Pexels

By Morgan Reyes

Before you read on: this guide covers a heart condition that can become an emergency. Read the scope and safety note first.

The Short Answer

As of July 2026 the FDA's investigation is still open, and it has never established a cause. Its last substantive public statement was 23 December 2022. There has been no recall and no ban.

What the evidence points at is not the absence of grain. It is how much pulse content a recipe carries and how high those pulses sit in the ingredient list. Both grain-free and grain-containing foods appear in the reports. Affected dogs generally have normal taurine, which rules out the simple deficiency that solved the equivalent problem in cats.

And the part worth knowing before anything else: caught early, this form often improves with veterinary treatment and a diet change. The genetic form does not.

The working rule: you do not need to bin the bag today, but you should look at where peas, lentils and potatoes sit on its ingredient list. Two published thresholds tell you what counts as too high.

What the FDA found, and what its own numbers can't tell you

The FDA received 1,382 reports of DCM in dogs between 1 January 2014 and 1 November 2022. That number is smaller and stranger than it looks.

In plain terms: a report means somebody told the FDA a dog had been diagnosed. It does not mean the agency checked, and it does not mean diet caused it. Those are three different claims, and the distance between them is most of this story.

Here is the agency's own breakdown, which almost no other page reproduces:

Three things in that table matter and rarely get mentioned.

Reports fell sharply at the end. 255 across 27 months, against 608 across 15 months in the window before. The FDA says so directly: it continues to receive reports "but at lower levels than experienced from 2018 through 2020."

The spikes track the FDA's own press releases rather than the disease. Again in the agency's words: "It is typical for FDA to receive a short-term increase in reports after issuing public updates on a pet health issue. In this case, upticks in the number of DCM reports to FDA tend to happen after FDA issues public updates."

There is no denominator. This is the most important caveat on the whole figure and it is almost never stated. The counts include "all reports in which one or more dogs was simply stated to have been diagnosed with DCM," the agency "is unable to investigate every report to verify or confirm the reported information," reporting is voluntary and "likely do not reflect the true incidence," and, plainly: "FDA is unaware of any statistics on the background rate of non-hereditary DCM in dogs." Without a background rate, 1,382 has nothing to be measured against.

One correction worth making out loud, because the wrong version is everywhere. You will often read that the FDA logged "over 1,100 cases between 2014 and 2022." That welds two separate things together: roughly 1,100 was the count as of about November 2020, published in a veterinary cardiology handout. The 2014-to-2022 total is 1,382. We bothered to check because an AI answer engine produced the wrong version, unprompted, while this guide was being revised.

Which dogs, and why the breed list is misleading

The breed most reported to the FDA is also the breed the FDA says is over-reported.

Golden Retrievers top the list at 95 reports, then mixed breeds at 62 and Labrador Retrievers at 47. And then the agency's own warning: it "has observed a reporting bias for breeds like Golden Retrievers due to breed-specific social media groups and activities that have raised awareness of the issue in these communities and urged owners and vets to submit reports to FDA."

So the top of that list partly measures which owners were paying attention.

Sitting a little lower are Great Danes at 25, Doberman Pinschers at 15 and Boxers at 11. Those breeds have a well-established genetic predisposition to DCM, and the FDA says what that implies: reported cases skewed male, "consistent with what is seen in genetic forms," and "it may be that some cases are genetic in origin or a combination of diet and genetic tendencies."

That is a regulator acknowledging its own case series is partly contaminated by ordinary hereditary disease.

One more line, to head off a common assumption: of the reported diets, 452 were dry food and 9 were raw. Whatever this is, it is not a raw-feeding story.

It was never really about grains

The FDA's own answer to "is this only a grain-free problem" is one word: no.

Asked directly, the agency replies: "No. FDA has received reports of non-hereditary DCM associated with both grain-free and grain-containing diets."

Here is the sentence that should change how you read a bag:

"Most of the diets associated with the reports of non-hereditary DCM have legume seed ingredients, also called 'pulses' (e.g., peas, lentils, etc.), high in their ingredient lists (although soy is a legume, we did not see a signal associated with this ingredient). These include both 'grain-free' and grain-containing formulations. Legumes, including pulse ingredients, have been used in pet foods for many years, with no evidence to indicate they are inherently dangerous."

Soy is a legume and it showed no signal. That one clause dismantles "legumes are bad for dog hearts" better than any amount of hedging, and almost no other page carries it.

Among the reported diets, 91% were labelled grain-free, 93% contained peas and/or lentils, 89% peas specifically, 62% lentils, and 42% potatoes or sweet potatoes. Those percentages cover 1 January 2014 to 30 April 2019 only, and were never recalculated for the larger 1,382-report set.

The FDA's laboratory work points the same way. Product testing "did not reveal any abnormalities" across minerals, metals and amino acids including taurine, and average protein, fat, taurine, cystine, methionine and resistant starch came out "similar for both grain-free labeled and grain-containing products." So this is not a straightforward nutrient shortfall, and that finding comes from the regulator's own lab rather than from a pet-food company.

The frame to carry away is pulse load and position, not the grain-free label.

What the research has and hasn't shown since the FDA went quiet

Since the FDA stopped publishing, more than 30 studies have appeared, and the two most important ones point in opposite directions.

In plain terms: one is independent and finds an association. The other is better designed, finds nothing, and was paid for by a pet-food company. You need both to understand where this sits.

Two things the table cannot carry.

The reassuring study is the best-designed evidence here, and you should know who paid for it. Randomized, double-blind, four diets, eighteen months, with echocardiography plus taurine plus cardiac biomarkers. It found nothing worrying. It was also funded and authored by Hill's Pet Nutrition, whose grain-inclusive products benefit from that conclusion. Its own authors report that no power analysis was done and that 60 dogs "may not be large enough to detect preclinical changes of DCM, which has a prevalence of approximately 0.5%." Twenty-four of the 60 developed mild endocarditis, which the authors themselves offer as the likely explanation for the echocardiographic changes they did see. And all four test diets were commercially formulated complete-and-balanced foods, which is the category nobody was worried about.

None of that makes it a bad study. It makes it a well-run study of a slightly different question, and several pages currently cite it as an all-clear without mentioning the funding at all.

The independent review disagrees with itself, and that is worth seeing. Its abstract says the findings "suggest a strong link between diet and DCM." Its conclusion says the evidence supports an "association." We are quoting both rather than picking whichever half reads better.

Cats, taurine, and why that story doesn't transfer

No. The cat version of this disease is solved and the dog version isn't, and the reason matters.

In cats, researchers found taurine deficiency causing dilated cardiomyopathy, showed that supplementing taurine restored heart function, and the industry reformulated cat food around it. Pion and colleagues published that in Science in 1987, studying 21 cats on commercial diets. Feline taurine-deficiency DCM became rare.

There are actually four different things sharing one disease name:

• Cats, taurine-deficiency DCM. Solved. Known nutrient, known fix.
• Dogs, a taurine-responsive subtype. Real but breed-linked. The FDA notes DCM "is also seen in Cocker Spaniels associated with taurine deficiency," and UC Davis documented it in Golden Retrievers.
• Dogs, non-hereditary DCM. The subject of this page. Affected dogs generally have normal taurine. Mechanism unknown.
• Dogs, hereditary DCM. Genetic. Not diet, and not reversible.

The asymmetry is this. In cats we found the missing nutrient before we understood the mechanism, and proved that replacing it reversed the disease. In dogs that nutrient has been largely ruled out for most cases, which leaves a suspected trigger with no established mechanism rather than a deficiency with a fix.

Our cat nutrition checklist covers the feline side. Read it as a contrast to this page, not as support for it.

What to actually do about the bag in your cupboard

Two published rules exist for where pulses may sit on a label. They disagree, and both are more useful than "low on the list."

The FDA's threshold is broader; the cardiology practice's is stricter. Neither is wrong. If you want one working rule, the stricter one costs nothing and rules out more.

Reading the list is a skill in itself, and it takes about a minute once you know what order to check things in. Our 60-second label guide walks through it, and what actually counts as a by-product covers the ingredient names that sound worse than they are.

Beyond the ingredient list, the World Small Animal Veterinary Association's criteria are about the company rather than the recipe: does it employ a qualified nutritionist, does it run feeding trials, does it own and control its production lines. Those questions survive a reformulation in a way a brand list does not.

If you have a Golden Retriever or an American Cocker Spaniel, both show up in the taurine-responsive subtype, and a taurine level is worth asking your vet about.

Do not abruptly switch food on your own. Both the FDA and UC Davis advise talking to a vet before choosing a boutique, grain-free or exotic-protein diet, and that cuts both ways: an unplanned overnight switch trades a hypothetical cardiac risk for a real digestive one.

Bottom line: the honest state of this question is an open investigation, a suspected mechanism nobody has pinned down, and one genuinely useful action. Look at where the pulses sit on your bag. If several are stacked into the first five ingredients, that is worth raising at your next appointment. It is not worth a panic, and it is not worth an unsupervised diet change.

FAQ

Is grain-free dog food banned or recalled?

No. There has never been a recall or a ban connected to this investigation. The FDA's position is that it "has no definitive information indicating that the diets are inherently unsafe and need to be removed from the market."

Open is not the same as concluded, in either direction.

Was my brand on the FDA list?

The FDA published 16 commonly reported brands in 2019 and then retired the list. Asked whether it would keep it updated: "No, we are not planning to update the commonly reported brands, as we are aware that several pet food companies have adjusted diet formulations since our initial announcements about DCM."

We are not reproducing a 2019 list, because a 2019 list tells you about 2019 recipes. The bag in front of you is better evidence.

Has the FDA said anything new since 2022?

No. The last substantive statement was 23 December 2022.

The Q&A page carries a footer reading "content current as of 08/19/2024," which looks like a later update. We compared an archived early-2024 capture against the current page: only navigation and table formatting changed. Every figure, date and answer of substance is identical.

How many dogs actually got sick?

1,382 reports between 1 January 2014 and 1 November 2022. That is not 1,382 confirmed diet-caused cases, for the reasons in the section above: reporting is voluntary, the FDA cannot verify every report, and there is no known background rate to compare against.

Should I switch my dog off grain-free right now?

Not abruptly, and not alone. If your dog is well, this is a next-appointment conversation.

Are peas bad for dogs?

Soy is a legume and showed no signal. The concern is pulse load and position in the ingredient list, not legumes as a category.

Can diet-associated DCM be reversed?

Often it improves if caught early, "unlike genetic forms of DCM" in the FDA's words, with veterinary treatment and a diet change. That contrast is exactly why early attention matters here.

Is this the same as the taurine problem in cats?

No. See the section above. Cats had a deficiency with a fix; dogs have a suspected trigger with no established mechanism.

A Note on Scope

This guide is information, not veterinary advice, and no veterinarian reviewed it. It cannot tell you whether your own dog has a heart problem. Talk to your vet before changing your dog's diet, and ask specifically about a boutique, grain-free or exotic-protein food before choosing one.

Signs that need a vet rather than a web page: coughing, laboured or fast breathing at rest, unusual weakness, or tiring far sooner than normal on a walk. Collapse or fainting means the same day. Dilated cardiomyopathy can be well advanced before anything is visible from the outside, which is precisely why the reversibility window described above depends on catching it early.

Sources

• FDA, investigation into a potential link between certain diets and canine dilated cardiomyopathy. The diet-pattern wording, the ingredient and breed charts, and the first-10-ingredients definition.
• FDA, questions and answers on its work on potential causes of non-hereditary DCM in dogs. The 1,382 total and its four-window breakdown, the background-rate admission, the retired brand list, and the reversibility contrast.
• AVMA, on the FDA ending public updates. Independent confirmation of the December 2022 stopping point.
• Mornard et al., "Role of Diet as a Predisposing Factor for Dilated Cardiomyopathy in Dogs: A Narrative Review," Veterinary Sciences 12(11):1106, 2025. No external funding, no declared conflicts.
• Morris et al., "Different carbohydrate sources in dog foods supported overall health and cardiac function," Journal of Animal Science, 2025. The 18-month randomized trial, funded by and authored at Hill's Pet Nutrition.
• Pion et al., "Myocardial failure in cats associated with low plasma taurine: a reversible cardiomyopathy," Science, 1987. The feline discovery this page is often confused with.
• Kaplan et al., "Taurine deficiency and dilated cardiomyopathy in golden retrievers fed commercial diets," PLOS ONE, 2018. The UC Davis taurine-responsive series.