Where vegans actually draw the line, and why the line moves

The Vegan Society's founding definition, tightened in 1979, says veganism seeks to exclude animal exploitation and cruelty as far as is possible and practicable. That clause is the load-bearing part of the sentence, because a diet and lifestyle with a zero-tolerance rule for every trace of animal involvement in a food system built on animal agriculture for ten thousand years would be unlivable, and the Society knew that when it wrote the definition. The interesting question is where different vegans, working from that same clause, end up putting the line.

This guide takes the hard cases in turn: where the argument for avoidance is strong, where it is genuinely contested, and where possible and practicable clearly means don't even try to avoid it. Each case links to the ingredient page that owns the verdict and the label detail. This page is about the reasoning that connects them.

Most of the disagreement below traces back to two ethical starting points that both call themselves vegan and reach different answers.

Rights-based abolitionism, the tradition closest to the Vegan Society's own language, treats animal use itself as the wrong, independent of whether any individual animal suffers. On this view, farming a bee for honey is wrong even if the bee is not in pain, because the bee is a being with interests that gets used as a resource. The test is not "does this cause harm" but "does this treat an animal as property."

Harm reduction, more common among the newer wave of people who arrive at plant-based eating through health or climate motives, treats suffering and death as the thing to minimize, and asks which choice causes less of it in practice. On this view, an egg from a rescued backyard hen who would lay the egg whether or not a human eats it is not obviously worse than a vegan diet's own downstream animal deaths from tilling, transport and pesticide use on plant monocultures. Both frames can point to real evidence and neither is a rationalization. They just answer "where is the line" differently because they are measuring different things.

Start with the case that has one right answer, because it is the case where confusion causes actual harm. UK law under the Human Medicines Regulations 2012 requires preclinical data, including animal testing, before a new medicine or vaccine can be licensed, and many tablets use gelatin capsules, lactose fillers or magnesium stearate that can be animal-sourced. The Vegan Society's own guidance on medicines says plainly that a person should discuss their beliefs and options with their healthcare professional, and that looking after your own health is part of being an effective advocate, not a betrayal of it. During the Covid-19 vaccine rollout the Society said outright that avoiding animal-tested vaccines was not possible or practicable at that point, and encouraged vegans to get vaccinated.

Oysters, mussels and clams have no brain and no central nervous system in the vertebrate sense. They have ganglia, clusters of nerve cells that let them respond to touch and close their shells, but nothing resembling the integrated pain processing that vertebrates have. A minority of people who otherwise eat a vegan diet, sometimes called ostrovegans, argue that eating an animal incapable of suffering does not violate the spirit of avoiding cruelty, only the letter of avoiding animal products. Most vegan organisations reject that math, on the reasoning that uncertainty about sentience should run in the animal's favor, not the eater's, and that bivalves are still animals under any zoological definition, which settles the label question even if it doesn't settle the suffering question.

Figs raise the same question from a stranger angle. A female fig wasp forces her way into an unripe fig to pollinate it and lay eggs, loses her wings and antennae doing it, and dies inside. The fig's own enzyme, ficin, breaks her body down into amino acids the fruit absorbs, so a wild-pollinated fig can genuinely contain the wasp's remains, digested. Common commercial fig varieties sold fresh or dried in the US and much of Europe are parthenocarpic and never need a wasp at all, which resolves the practical question for most shoppers but not the philosophical one: does a natural mutualism between two wild species, with no human intent behind the wasp's death, count as the kind of exploitation the definition is about. The ingredient page carries the varietal detail; see /is-it-vegan/figs/.

Palm oil is the case that best separates the two ethical frames, because it is entirely plant-derived and fails no ingredient test, yet its expansion has been one of the most cited drivers of habitat loss for orangutans, tigers and elephants across Sumatra and Borneo, with satellite mapping in 2024 and 2025 still showing plantation clearing inside protected reserves. A rights-based vegan asks whether the oil itself came from an animal, and it did not, so the ingredient passes. A harm-reduction vegan asks whether buying the product funds a supply chain that kills animals at scale, and by that test palm oil fails harder than many things vegans do avoid. Neither the Vegan Society nor Certified Vegan disqualifies palm oil from certification, which tracks the rights-based reading of the definition, but a serious number of vegans avoid it anyway on harm grounds and buy RSPO-certified or palm-free products instead. Full sourcing detail is at /is-it-vegan/palm-oil/.

The same split explains honey. A honeybee is unambiguously an animal being managed for a product, which settles it for a rights-based vegan without much debate. A harm-reduction vegan sometimes points out that a well-run apiary, unlike a battery farm, does not confine or slaughter its bees for the product, and asks whether that changes the calculus. The Vegan Society's position, and the mainstream vegan position, is that management and use of a sentient being for what it produces is exploitation regardless of how comfortable the arrangement is for the bee, so honey does not pass. See /is-it-vegan/honey/ for the beekeeping detail.

Backyard hens split the vegan community more visibly than almost anything else on this list, to the point that some vegan forums ban the topic outright because the argument never resolves. The harm-reduction case: a hen rescued from a factory farm or a friend's smallholding is going to lay eggs whether or not anyone eats them, the household isn't paying anyone to breed or confine her, and letting the eggs go to waste doesn't undo any exploitation that already happened. The rights-based case: taking and eating an egg still treats the hen's biological output as a resource for humans, egg-laying at modern hybrid-hen volumes shortens a bird's lifespan through calcium depletion regardless of who eats the result, and the exception has a way of sliding into justifying eggs from less clearly rescued birds. Both sides agree the hens should be rescued and cared for. They disagree about what happens to the eggs, and neither the Vegan Society's definition nor its certification scheme resolves it, because it is a question about what "use" means, not a question about ingredients.

The shared feature of bone char and finings is that the animal material is a processing aid, not an ingredient: it does its job during manufacture and is meant to be removed before the product reaches a shelf. Rights-based vegans still avoid these when a supplier confirms their use, on the reasoning that paying for a process that uses an animal, even one that leaves before the final product, is still funding animal use, and that traces do sometimes carry through. Harm-reduction vegans are more likely to shrug at a product where testing shows no detectable animal material remains. Brand-level detail, including which major sugar refiners and wineries use which method, lives at /is-it-vegan/bone-char-sugar/, /is-it-vegan/isinglass/ and /is-it-vegan/wine/.

Car tyres commonly use stearic acid to help vulcanized rubber hold its shape and grip the road, and most stearic acid on the market is rendered from animal fat because it is cheaper than the plant-derived version, though chemically identical once refined. Michelin is the tyre maker most often cited as using plant-derived stearic acid across its full range; several other major brands, including Kenda, Continental and Goodyear, offer some vegan lines without converting entirely. Road surfaces themselves sometimes use bitumen extenders that include animal fat derivatives, though this varies by supplier and is harder for a driver to check than a tyre spec sheet. See /is-it-vegan/car-tyres/ for the brand-by-brand detail.

UK banknotes are the cleanest illustration of "avoidable but not avoided." The Bank of England confirmed in 2016 that its polymer notes contained trace tallow, added to help notes run smoothly through counting and vending machines. After a public petition drew more than 130,000 signatures, the Bank ran a consultation and in August 2017 decided to keep the tallow-based formulation, citing the added cost, estimated near £16.5 million over a decade, and environmental tradeoffs of the palm-based alternative it had tested. A vegan in the UK who buys nothing but produce with cash is still, structurally, transacting in tallow, and the near-universal vegan response is that this is precisely the kind of unavoidable background exposure the "practicable" clause exists to cover: you cannot opt out of the national currency. Detail at /is-it-vegan/bank-notes/.

Cats are obligate carnivores. They cannot synthesize enough taurine, an amino acid, from plant precursors, and a taurine-deficient diet in cats causes dilated cardiomyopathy and retinal degeneration, so any vegan cat food has to add synthetic taurine and several other nutrients that would otherwise come from meat. Some formulations, tested by manufacturers, have kept cats at normal taurine levels; independent veterinary opinion is split on whether that success generalizes reliably across brands and individual cats, and most veterinary nutrition bodies still recommend a meat-inclusive diet for cats specifically, while treating dogs, who are flexible omnivores, as a separate and less fraught case. The ethical tension here does not resolve the way the others do, because the being whose interests are at stake is the pet, not the vegan choosing its food, and the practicability clause was written with the vegan's own choices in mind, not a decision made on a dependent animal's behalf. Vegans who keep cats mostly split between conventional meat-based cat food chosen on welfare and sourcing criteria, and closely vet-monitored vegan formulations; almost none treat the choice as obvious.

A pattern runs through every case above. Where an animal is bred, confined, milked, mined for a substance, or killed to order for a product, and an alternative exists, vegans converge: that's honey, that's isinglass wine, that's bone char sugar where an alternative is on the shelf. Where the animal involvement is a byproduct of infrastructure nobody individually controls, like currency or road surfaces, vegans converge the other way and call it impracticable to avoid. The genuine disputes, bivalves, figs, palm oil, rescued eggs, sit in the middle, where reasonable people applying the same definition weigh direct causation against funded harm differently. That is not the definition failing. A standard that bites cleanly on the easy cases and produces honest disagreement on the hard ones is doing what a workable ethical standard is supposed to do. The alternative, a rule with no give in it at all, would either be unlivable or would quietly stop being followed, and neither outcome protects animals better than an honest, case-by-case line does.