How better ways of screening drugs could soon lead to thousands fewer animal tests

As many as 190 million animals are used in biomedical research around the world every year. But after centuries of experiments, controversy and countless life-saving benefits, science may finally be approaching a point where animal testing is no longer the default option. Until recently, that idea would have sounded implausible. Modern biomedical research has relied on animals almost since its inception, using them to study disease and test new medicines before they ever reach people. But over the past few years, that assumption has begun to shift. Efforts to replace animals in science have the firmest backing yet from policy-makers and regulators. In April 2025, the US Food and Drug Administration (FDA) published a five-year road map to reduce animal testing in drug-safety studies. Months later, the country’s National Institutes of Health (NIH) stopped requiring animal models for many of its grants. Late last year, the UK launched a similar plan to phase out animal testing, and the European Union followed suit in June. Advertisement These initiatives largely seek to make more widespread use of better alternative methods. A sliver of plastic no bigger than a microscope slide threaded with human liver cells can now identify toxic drug effects better than using rats. Lab-grown cells are sparing rabbits from once-indispensable tests that screen medicines for contaminants. And AI models are identifying potential drug targets at record speeds, without using animals. “The science has reached a point where there is a much better evidence base to support the fact that phasing out much animal testing might be possible,” says Celean Camp, head of the UK campaign group Replacing Animal Research. The ancient history of animal testing Animal experimentation dates at least as far back as the ancient Greek philosopher Aristotle, who dissected dozens of animal species to better understand anatomy. But modern systematic use of animal experiments didn’t start until the mid-1800s. Then, and now, advocates for using animals argued that many of them have bodies similar enough to humans to justify tests on biological mechanisms and drug effects, thus informing science and improving healthcare without injuring people. Pushback started almost immediately. Public horror at some of the Victorian physiologists’ methods prompted the UK to pass the world’s first animal testing law in 1876, which ensured it was done only by licensed scientists. Yet, as generations of scientists have repeatedly said, nobody really wants to run animal experiments. They are slow, expensive and come with significant paperwork – and that is before everyone involved weighs up the ethics of deliberately inflicting pain on animals, including dogs, pigs, monkeys and, yes, guinea pigs. But, as distasteful as it might be to some, animal testing has let researchers probe biological processes and screen experimental medicines. It has facilitated monumental breakthroughs, from the discovery of insulin and antibiotics to the development of pacemakers and covid-19 vaccines. Most modern regulations on animal testing rest on a 1959 concept known as the 3Rs: replacement, reduction and refinement. “Replacement” means replacing animals with computer models or bacteria where possible; “reduction” is using as few animals as possible; and “refinement” is minimizing pain and distress. Countries often vary in how they apply the framework. Rules in the UK and the EU, for example, cover all vertebrates, but the US excludes mice, rats and birds bred for research, which constitute the majority of test animals. But there are still a lot of animals used in research (see box below). About half of animals used in scientific tests these days are involved in basic research – the kind done at universities to uncover how biology functions and fails. The other half is split roughly between testing new drugs for side effects and setting safe exposure limits for chemicals, such as pesticides. With some exceptions, public and political opinion has largely tolerated that trade-off. But there are signs of waning support. In 2001, most Americans believed medical testing on animals was morally acceptable. By 2025, fewer than half did. The most recent UK survey, from 2018, found support holding steadier at about two-thirds – but also a growing appetite for alternatives, especially among younger people. “The public has always been somewhat supportive of animal research where there is no alternative,” says Camp. But new techniques now lessen the need for an increasing number of animal tests, and scientists haven’t caught up with that, she says. “There’s more and more evidence that the regulatory mechanisms to check that there is no alternative have not been functioning as well as they could,” says Camp. “So that assurance on which the public trust and the public support stands is a little shaky.” Regulatory agencies are now trying to accelerate the transition to non-animal methods – and have been for a while. “The whole push towards more non-animal models isn’t recent,” says Ivan Rusyn, a toxicologist at Texas A&M University. Instead, he says, the recent proposals result from a “perfect storm” of political, scientific and regulatory activity over more than a decade. For example, back in 2013 the EU banned animal testing on cosmetics, proving regulators could force the issue and that industry would adapt. A new world of testing possibilities The recent UK and EU road maps aim to build on this progress and develop more non-animal regulatory tests. The US approach is slightly different and aims to replace animals in scenarios where they are notoriously poor models – such as for testing monoclonal antibodies, a treatment that mimics the immune system. Neither transition intends to be a wholesale replacement, as none of the alternatives can fully recreate a whole-bodied animal. But used in the right context – replacing an experimental method here or a safety check there – the technologies can meaningfully reduce animal testing and even produce better results. So, what actually are the alternatives? One of the most established is cell cultures, which let researchers expose living human tissue to chemicals. These can replace some mandated drug and chemical toxicity tests, but a flat dish of cells is a poor substitute for the changing environment of a working organ. Devices called organs-on-a-chip try to close that gap by recreating some of that dynamism. Roughly the size of a USB stick, they house cells on a flexible scaffold lined with tiny fluid-filled channels that supply nutrients and remove waste. Organs-on-a-chip have a significant leg up on animal models when it comes to testing drug safety. More than 90 per cent of experimental drugs that appear safe and effective in animals don’t receive regulatory approval, largely because they turn out to be toxic to humans. In 2022, biologist Donald Ingber at Harvard University and his colleagues at the biotech company Emulate, which he co-founded, built and tested the toxicity of experimental drugs using a liver-on-a-chip. It identified nearly 7 out of every 8 drugs that proved toxic in clinical trials despite appearing safe in animal tests. “People are ignorant about how far this field has progressed,” says Ingber. The chip, which is now in the final stages of FDA approval, has already made its way into labs. The pharmaceutical company Moderna has used it in place of primates to screen compounds used in mRNA vaccines for liver damage. Ingber’s lab also builds chips to mimic other organs, including the cervix, lungs, kidneys and intestines, and wires them together with artificial blood vessels. “We have, like, a human body-on-chips,” he says. Mounted on a scaffold, they can physically flex to mimic breathing or gut motion, addressing one of the oldest complaints about cell cultures: that real tissues never work in isolation. Even more complex are organoids, self-assembling blobs of human cells that mimic organs. Rather than growing cells on an engineered scaffold, as chip-builders do, organoid researchers stimulate stem cells – which can transform into other cell types – into miniature, three-dimensional tissues. The resulting “mini organs” capture more of a tissue’s structure and cell variety. Hans Clevers at Utrecht University in the Netherlands, a molecular geneticist who pioneered organoids, argues the technology can already outperform animal models for some human conditions, including cystic fibrosis. “We have animals that beautifully mimic the symptoms, but they typically don’t have the underlying cause,” he says. That distinction matters because laboratory animals have always represented a compromise. Scientists can coax mice to develop cancers, inflammatory diseases and neurological disorders resembling human illness, but such conditions often arise through genetic engineering or artificial intervention, rather than the biological processes that make people sick. Clevers says his team developed an antibody to treat a head and neck cancer called squamous cell carcinoma entirely in organoids, with no conventional animal testing. The drug is now in phase III trials. Organoids and organs-on-a-chip also open a door to personalised medicine in a way that animals can’t. Researchers can grow them using a patient’s own cells and then test candidate drugs directly on tissue sharing that patient’s biology. Other animal alternatives don’t involve biological material at all; they run on computers in what researchers call in silico models. AI systems trained on genetic, chemical and clinical datasets can flag potential drug targets and predict toxicity. One problem, though, is that models don’t always explain how they generate decisions, which makes regulators cautious about relying on them to judge safety. Molecular docking software goes further and simulates how a potential drug physically fits and binds to its target protein, letting chemists screen thousands of compounds on a laptop. Then there are pharmacokinetic models, which divide the body into mathematical compartments – liver, kidneys, bloodstream – to predict how they absorb, break down and clear drugs. The approach could help estimate safe doses in groups of people rarely included in trials, such as children or pregnant women. None of these tools alone gives the complete picture of a drug. But together, they can help researchers zero in on compounds most likely to succeed and then investigate them further with organoids, organs-on-a-chip and remaining animal tests. Obstacles to accepting the new science In practice, it can still take years, or even decades, for non-animal methods to be widely adopted. Take fever tests, a widely used way of checking whether drugs contain contaminants. For more than a century, regulators have mandated injecting new drugs into rabbits to see if they cause the animals to develop a high temperature, signalling an infection – simple, effective and unpleasant for the rabbit. Alternatives to using rabbits have existed for decades. For example, a screening test using horseshoe crab blood, which clots on contact with bacterial toxins, dates to the 1970s. Then, a non-animal test emerged in the 1990s: a human blood test that measures immune responses to contaminants. Yet, it took until this year for the rabbit test to be officially phased out, and then still only in Europe – it is ongoing in some countries, including the US. Some of the lag comes down to understandable caution. “The regulators say they can only make judgements on whether the data is adequate once people submit it to them,” says Camp. “And the companies are saying, well, we’re only going to submit data that we know the regulators are going to accept.” The new road maps are attempting to break this impasse by validating non-animal models and creating guidelines around their use in drug development. It can also be challenging for career scientists to move from familiar territory. “Academics generally tend to do incremental stuff, and it’s very hard to learn a whole new methodology,” says Ingber. There are some signs of movement, however. A 2024 analysis of research on breast cancer, lung disease, blood cancer, heart disease, neurodegenerative disease, diabetes and toxicology found that by 2022, publications using only non-animal methods had overtaken those using animals in all seven areas. That could be due to a rise in tests developed as a response to the EU cosmetic-testing ban. Government money may also accelerate the shift. China launched a project in 2024 to speed up the development of lab-grown human tissue models. In the US in 2025, the NIH opened its first organoid-development centre, while halting funding for research using only animal models. In Australia, the New South Wales government allocated $4.5 million in 2024 for a programme to reduce and replace animals in medical research. None of this is to say that animal research will ever be phased out entirely. “It’s animal research that gives us an understanding at the organism level that all of this in-vitro stuff, including organoids, do not,” says molecular and cell biologist Michael Hadjiargyrou at the New York Institute of Technology. Delete a gene from cells in a dish, and the consequences are often minor. Delete it from a living animal, and it can reshape their physiology. “We need to educate Congress. They need to know how critical animal research is to scientific discovery,” says Hadjiargyrou. Rusyn thinks political rhetoric on reducing animal testing has occasionally run further than the evidence justifies. He is particularly unimpressed by a US Environmental Protection Agency’s pledge to drop mammal testing requirements for pesticides by 2035. “It is completely bonkers, in my opinion, as a toxicologist,” he says. Because cell cultures are often more sensitive to chemicals than an intact animal, they risk banning substances that would otherwise be harmless at the doses people encounter. Even committed opponents like Camp accept that some animal research will always be necessary. It is difficult to see how research into psychiatric and neurodegenerative diseases, for example, could proceed otherwise, given the complexity of the central nervous system and the brain. But there is some evidence that the regulatory drive for replacements is having a positive impact. In its first-year progress report, the FDA announced that a change to primate toxicity testing for certain immune-based therapies had saved hundreds of non-human primates and that there is “potential for [saving] thousands more animals as methods expand”. But while the shift to alternative methods may be a long and laborious process, there are at least 190 million reasons to keep trudging ahead.