Can dogs and cats be vegan? The science says these are two different questions

In 1987 a research team led by veterinary cardiologist Paul Pion found that cats fed ordinary commercial cat food were dying of heart failure caused by a nutrient deficiency nobody had been tracking. The nutrient was taurine, an amino sulfonic acid barely present in plants, and the finding forced every pet food maker in the country to reformulate. That history matters here because it shows the failure mode is not hypothetical and it did not require anyone to try feeding a cat vegan. It happened with meat already in the bowl. Ask whether dogs and cats can go vegan and you are really asking two unrelated questions, because dogs and cats digest, metabolise and synthesise nutrients differently enough that a diet reasoned out for one tells you almost nothing about the other.

Dogs are classified as facultative carnivores, meaning their digestive system can run well on a mixed or largely plant diet, and the clearest evidence is genetic rather than anecdotal. Erik Axelsson and colleagues sequenced dog and wolf genomes in 2013 and found that dogs carry, on average, about seven times as many copies of the AMY2B gene, which codes for pancreatic amylase, the enzyme that breaks starch into digestible sugar, and that the same study found parallel selection on MGAM and SGLT1, the genes that finish converting starch to glucose and transport it across the gut wall. A follow-up study by Arendt and colleagues, published in Animal Genetics in 2014, linked that copy number variation directly to enzyme output, finding roughly a 5.4 percent rise in blood amylase activity for each extra AMY2B copy. Later work on ancient European dog remains, published in Royal Society Open Science, dated much of this expansion to the spread of agriculture, when dogs living alongside farming humans ate a lot more grain than dogs living alongside hunter-gatherers. A wolf did not evolve this; a dog living near a granary did.

That adaptation explains why a dog can extract real nutrition from starch a wolf would mostly waste, but it does not by itself make a vegan diet nutritionally complete. Dogs still have specific requirements for taurine (though, unlike cats, they synthesise it well from methionine and cysteine in most circumstances), for essential amino acids like lysine and methionine that are less concentrated in plant proteins, and for long-chain omega-3s that come almost entirely from animal or algal sources in nature. A vegan dog food has to be formulated to cover these gaps deliberately, not simply substituted ingredient for ingredient.

The largest single dataset on real-world vegan dogs comes from veterinary academic Andrew Knight, who surveyed 2,536 dog guardians and published the results in PLOS ONE in 2022. Dogs on vegan diets for at least a year showed lower rates of guardian-and-vet-reported illness across seven general health indicators, with reductions of 14 to 51 percent against conventionally fed dogs, and lower rates of six specific disorders. That is a genuinely striking result and it gets cited constantly.

It also has real methodological weight against it. The data is guardian-reported, not clinically measured: nobody ran bloodwork, nobody had a vet examine every dog by a common protocol, and "reported unwell" bends toward whatever an owner already believes about their dog's diet. The sample was self-selected: an owner who chose to feed a vegan diet and volunteered for a plant-based diet survey is, on average, more invested in that diet succeeding than a random dog owner would be, which tends to inflate reported satisfaction on any measure. Knight himself follows a vegan diet, and the study disclosed funding from ProVeg International, an organisation that advocates for reducing animal product consumption. None of that makes the numbers fabricated. It means the study establishes that vegan dogs can look healthy by guardian report under real-world conditions, and it does not establish causation, blind assessment or clinical endpoints. Treat it as a genuinely useful data point sitting well below the evidentiary bar of a controlled feeding trial, because it is one.

Cats are obligate carnivores, a term with a specific metabolic meaning: their bodies evolved several enzyme systems down to near-zero activity because a wild feline diet supplied those nutrients pre-made, every time, from prey tissue. Removing meat does not just remove a food group for a cat. It removes the only route to several compounds its own cells cannot manufacture in adequate quantity.

Taurine is the clearest case, and it hinges on a single enzyme. Producing taurine from the amino acids cysteine and methionine requires cysteine sulfinic acid decarboxylase, and cats have very low activity of that enzyme relative to dogs, compounded by a competing enzyme, aspartate aminotransferase, that diverts the same precursor down a different pathway before it can become taurine. On top of that, cats use taurine exclusively, not glycine like most mammals, to conjugate bile acids for fat digestion, and they excrete a portion of it with every digestive cycle regardless of intake. The result is an obligatory dietary requirement that a cat cannot economise its way around.

Vitamin A is the second case and the mechanism is different: plants supply beta-carotene, and most mammals run an enzyme, beta-carotene 15,15'-monooxygenase, that splits it into usable retinol. Cats have essentially no useful activity of that enzyme, so beta-carotene passes through a cat largely unconverted, meaning carrots and sweet potato do nothing for a cat's vitamin A status no matter the quantity. The third case is arachidonic acid, an omega-6 fat central to a cat's skin, kidney and reproductive function. Most mammals convert it from linoleic acid, found in plant oils, using delta-6 desaturase; cats have limited activity of that enzyme too, so arachidonic acid functions as close to essential for a cat as a nutrient gets from diet alone.

In 1987, veterinary cardiologist Paul Pion and colleagues at UC Davis published a study in Science that found 22 of 23 cats diagnosed with dilated cardiomyopathy, a condition in which the heart chambers enlarge and lose their ability to pump, had low plasma taurine, and that plasma taurine correlated with echocardiographic heart function across the group studied. When cats in the study were switched to a taurine-supplemented diet, most saw their heart function normalise within months. Separately, taurine deficiency causes central retinal degeneration, a slow, progressive death of photoreceptor cells that produces blind spots and can progress to full blindness, and it is not reliably reversible once damage has occurred, unlike the cardiac form. The 1987 finding is why every AAFCO-compliant cat food today carries a taurine minimum, and why the entire pet food industry treats taurine testing as routine. It is worth sitting with the fact that this crisis happened inside conventional, meat-based commercial cat food. The industry did not know it needed to test for this until cats started dying of it.

Andrew Knight has also studied vegan cats specifically, publishing guardian-reported outcomes in a 2022 PLOS ONE paper with a broadly similar shape and the same methodological caveats: self-selected owners, unblinded outcome reporting, no clinical bloodwork. It is not nothing, but it cannot answer the specific question that matters for cats, which is whether the diet in the bowl actually delivers bioavailable taurine, arachidonic acid and retinol at the dose a given cat's body needs, day after day, for years.

That question has been tested directly, and the results are uncomfortable for anyone hoping formulation alone solves the problem. A 2020 PLOS ONE study of commercial vegan foods marketed for dogs and cats in Brazil found every product tested fell short of at least one recommended nutrient level, with calcium, phosphorus and the calcium-to-phosphorus ratio the most common failures, and one product's cat food came up short on potassium. A 2024 systematic review pooling 54 separate nutrient-adequacy studies of pet foods found only 18.5 percent came back with no deficiencies or excesses against AAFCO, FEDIAF or NRC guidelines at all, across every diet type studied, and taurine deficiency turned up specifically and repeatedly in cat foods. One study cited in that review tested 31 commercial foods and found zero passed every nutritional check; 55 percent met amino acid targets and only 16 percent met mineral targets. This is not evidence that vegan cat food is uniquely bad. It is evidence that formulating a complete diet on paper and actually delivering that nutrition in a manufactured, shelf-stable product are different engineering problems, and pet food as a category has a track record of getting the second one wrong even for supposedly straightforward meat-based recipes.

"Complete and balanced" is not marketing language, it is a regulatory claim with two possible legal bases in the United States, and they are not equivalent. A food labelled as formulated to meet AAFCO's nutrient profiles was designed on paper, using a database calculation of the recipe's expected nutrient content, and never had to prove that a real animal absorbs those nutrients at the stated rate. A food labelled as having passed AAFCO feeding trials was actually fed to a group of animals for a defined period, typically around six months, under a monitored protocol, with bloodwork and weight checks along the way. Europe's FEDIAF standard, the equivalent body across the EU, runs on the same underlying logic, calculation-based formulation as the default, feeding trials as the stronger and less common claim. A cat food that says "formulated to meet" is telling you a spreadsheet balanced, not that a cat's body actually took up the taurine on the ingredient panel; bioavailability, the amino acid interactions the Brazilian study flagged, and manufacturing losses from heat processing can all still leave a gap between the label and the bowl.

Between 2018 and 2019 the FDA investigated a cluster of dilated cardiomyopathy cases in dogs, many eating grain-free diets heavy in peas, lentils and other legumes, reviewing 524 reports covering 560 dogs and 14 cats by mid-2019. Some of the affected breeds, including Golden Retrievers, had no known genetic predisposition to DCM, which is what triggered the alarm. The agency kept the investigation open for years afterward: measured protein, fat, taurine and methionine-cystine levels came back similar between grain-free and grain-containing products on a dry-matter basis, and no single ingredient or mechanism was ever confirmed. In December 2022 the FDA stated that it had not found evidence linking grain-free or boutique, exotic-ingredient, grain-free diets to the DCM cases, and that it would not issue further public updates absent meaningful new scientific information, closing out a multi-year investigation that never identified a cause. Legumes were never shown to cause the heart disease, so a vegan or plant-heavy pet diet does not inherit a proven risk from this history. What the investigation does show is that a confident nutritional claim on a pet food bag, formulated by people who genuinely believed they had covered the requirements, can still coincide with real, serious, hard-to-explain disease in the animals eating it, years into a lengthy federal review, without ever being pinned to a mechanism. Confidence at the point of formulation has repeatedly proven not to be the same thing as safety at the point of years of consumption.

For dogs, a vegan diet is biologically plausible and has real, if imperfect, supporting evidence: the AMY2B expansion gives dogs genuine digestive latitude that wolves lack, and Knight's survey data, weak as it is on causation, at minimum shows large numbers of vegan dogs are not obviously suffering under real-world guardian care. The honest caveat is that plausible is not the same as easy. Formulation has to specifically address amino acid balance and long-chain omega-3s, and periodic veterinary monitoring is the difference between plausible and reckless.

For cats, the burden of proof sits much higher, because the failure modes are severe, the taurine and vitamin A pathways are shut almost completely rather than merely reduced, and the 1987 discovery shows how long a deficiency can hide inside a diet everyone assumed was fine. Vegan advocacy itself typically includes a practicability clause, the idea that veganism means avoiding animal exploitation as far as is possible and practicable, and for an obligate carnivore that clause reasonably points toward the cat's physiological needs rather than the owner's dietary preference. That is not a verdict against vegans owning cats. It is a case for feeding the cat what its biochemistry actually requires, monitoring it properly if a plant-based formulation is used at all, and being honest that the standard of evidence a cat's health deserves is higher than an owner-reported survey can currently provide. The where vegans draw the line guide covers this specific edge case within the wider ethics of pet keeping.