What Has Animal Testing Achieved for Modern Medicine?

Animal testing has contributed to virtually every major class of medical treatment used today, from insulin and anesthetics to heart surgery techniques, vaccines, and cancer immunotherapies. That does not mean every animal experiment has translated cleanly into human medicine. The record is a mix of genuine breakthroughs and significant translational failures, and the field is now in the middle of a slow but real shift toward human-centric alternatives. Understanding what animal research has actually delivered, and where it has fallen short, gives a clearer picture than either uncritical celebration or blanket dismissal.

Insulin and the Experiments That Changed Diabetes

One of the most cited achievements of animal testing is the discovery of insulin. The link between the pancreas and diabetes was established in the late 1800s, when researchers removed the pancreas of a dog and observed that it developed uncontrolled urination and sugar in the urine. When they grafted a piece of pancreas back into the animal, the symptoms disappeared. Removing the graft again brought diabetes back. That back-and-forth experiment provided definitive proof that the pancreas secreted something critical for controlling blood sugar.¹

Decades later, in 1921, Frederick Banting and Charles Best at the University of Toronto built on that foundation. They used dogs to test whether ligating the pancreatic ducts before extracting the organ would preserve the blood-sugar-regulating substance while destroying the enzyme-secreting tissue. Their extract lowered blood sugar in dogs that had been made diabetic by removing their pancreases. The team then developed a way to extract the substance from whole beef pancreas without the ligation step, and that extract worked in humans with diabetes.²

Before insulin, a diagnosis of type 1 diabetes was essentially a death sentence. The drug’s introduction in the early 1920s turned it into a manageable chronic condition. It remains one of the clearest examples of an animal-derived discovery that directly saved millions of human lives, and the core approach of using pancreatic extracts came entirely from experiments in dogs.

Open-Heart Surgery and the Two Decades of Animal Work Behind It

Modern cardiac surgery owes much of its existence to animal research, particularly in dogs. In 1931, a young surgeon named John Gibbon began animal experiments in Boston aimed at building a machine that could temporarily take over the functions of the heart and lungs. It took him nearly twenty years. In May 1953, he performed the first successful open-heart operation using a heart-lung machine in Philadelphia.³

The cardiopulmonary bypass was just one of many cardiac advances that depended on canine models. Research in dogs also enabled the use of hypothermia during surgery, the development of operations for congenital heart defects, heart valve replacement, heart transplantation, and surgical treatments for abnormal heart rhythms.⁴ The Blalock-Thomas-Taussig shunt, a procedure that saved the lives of babies born with a heart defect that starved their blood of oxygen, was developed and perfected in dog models before it was ever tried on a child. Without those decades of iterative animal work, surgeons would have had no way to learn how to stop a human heart, repair it, and restart it safely.

Anesthesia and Pain Management

Animal testing also underpins the drugs that make surgery survivable in the first place. The properties of nitrous oxide, ether, and chloroform were all tested in animals before anyone attempted to use them on a human patient. Intravenous anesthetics like barbiturates and propofol, local anesthetics like cocaine, and neuromuscular blocking agents all went through animal testing as a critical step in development.⁵ Understanding how these drugs affect breathing, heart rate, and consciousness required studying their effects in living organisms before risking human lives. Much of what anesthesiologists know about monitoring physiological systems during surgery traces back to animal studies of these drugs.

From Polio to COVID Vaccines

Vaccine development has relied on animal models for over a century. The polio vaccine, which eradicated a disease that paralyzed tens of thousands of children each year, was developed and safety-tested in primates. More recently, animal studies played a visible role in the rapid development of COVID-19 vaccines.

The Moderna mRNA vaccine (mRNA-1273) was tested extensively in nonhuman primates before entering human trials. In those studies, vaccinated animals developed antibody levels that exceeded those found in humans who had recovered from COVID naturally. At the higher dose, viral replication was undetectable in the nose of all eight animals by two days after being deliberately infected, and the lungs showed limited inflammation or detectable virus.⁶ Follow-up work showed that the vaccine’s protection was driven by antibody responses in a dose-dependent way, and that transferring antibodies from vaccinated primates to unvaccinated hamsters was enough to protect the hamsters from infection.⁷

Other COVID vaccine candidates went through similar preclinical testing. One mRNA candidate elicited strong neutralizing antibodies across mice, hamsters, and primates, and protected hamsters from the weight loss and lung damage caused by infection.⁸ These animal studies mattered because they provided early evidence that the vaccines generated the right kind of immune response and did not trigger dangerous inflammation, which helped justify moving into human trials at unprecedented speed.

Deep Brain Stimulation and Parkinson’s Disease

Parkinson’s disease treatment provides a particularly clear example of how animal research led to a therapy now used in hundreds of thousands of patients. The key breakthrough came from studies in monkeys using a neurotoxin called MPTP, which produces symptoms closely resembling Parkinson’s disease. Researchers used this monkey model to map how the disease disrupts brain circuits and to build models of what goes wrong in the clusters of brain structures that control movement.⁹

That work provided the scientific rationale for reviving a surgical approach called pallidotomy and, eventually, for developing deep brain stimulation, a procedure in which electrodes implanted in the brain deliver electrical pulses that reduce tremors and stiffness. Without the monkey model, researchers would not have understood which brain regions to target or why electrical stimulation of those regions would help. Deep brain stimulation is now a standard treatment for advanced Parkinson’s disease and is being explored for depression, obsessive-compulsive disorder, and other conditions.

Cancer Immunotherapy and Mouse Models

The revolution in cancer immunotherapy, particularly the success of immune checkpoint drugs that unleash the body’s own immune system against tumors, was built on decades of research in mice. Mouse cancer models have been the primary platform for testing whether blocking specific immune checkpoints can shrink tumors.¹⁰ More recently, researchers have developed “humanized” mice, which are immunodeficient mice transplanted with human tumors and human immune cells. These models allow scientists to study how human-specific immunotherapies interact with human tumors in a living system, something no simple cell culture can replicate.¹¹

The honest caveat is that mouse immune systems differ enough from human ones that plenty of promising mouse results have not panned out in people. Researchers acknowledge that the number of preclinical models that faithfully capture the human immune system’s interaction with tumors is still limited. But the humanized mouse approach has made testing more realistic, and it remains one of the few ways to study combination immunotherapy strategies before trying them in patients.

Knockout Mice and the Genetics of Disease

Beyond testing drugs, genetically modified mice have become indispensable for understanding what individual genes do. By “knocking out” a single gene and observing what goes wrong, researchers can identify which genes cause which diseases. This approach has been especially valuable for studying rare diseases, where knockout mice provide evidence linking specific gene deletions to problems across multiple organs.¹² For conditions that affect only a handful of people worldwide, mouse models are sometimes the only feasible way to study the disease at all.

More broadly, the ability to create transgenic and knockout mice has reshaped how researchers connect genetic profiles to disease. These models allow scientists to watch a disease develop from its earliest molecular stages and to test potential therapies on animals with a known genetic makeup.¹³ In neuroscience, knockout and transgenic mice have been particularly useful for studying complex problems involving brain function and behavior, where the traditional approach of simply observing symptoms was too blunt.¹⁴

Where the Track Record Gets Complicated

For all the genuine achievements, the translation rate from animal studies to successful human treatments is strikingly poor. The failure rate for drugs moving from animal testing to approved human treatments remains above 92%, and it has been stuck at roughly that level for decades. Most failures are due to unexpected toxicity in humans that was not apparent in animal tests, or to drugs that simply did not work in people despite working in animals.¹⁵

The reasons go deeper than bad luck. Mice and humans differ substantially in which genes are turned on during inflammatory responses, and even in the timing and duration of gene expression during disease. In one telling example, when a human gene that produces a protein involved in blood sugar control was expressed in genetically altered mice, it had the opposite effect from what it does in humans: it caused loss of blood sugar control in the mice.¹⁶ Even when an animal model faithfully mimics a human disease, differences in physiology, behavior, and how drugs are processed in the body can undermine the results.

A systematic review comparing treatment effects in animal experiments with the results of human clinical trials illustrates the inconsistency. Corticosteroids appeared to help in animal models of head injury but showed no benefit in human trials. Tirilazad, a drug for ischemic stroke, reduced brain damage and improved function scores in animal models but was associated with worse outcomes in patients. On the other hand, thrombolysis for stroke and bisphosphonates for osteoporosis showed reasonably consistent effects across animal models and human trials.¹⁷ The picture is not that animal models are useless; it is that their predictive accuracy is uneven and sometimes dangerously misleading.

A review of systematic assessments of animal experiments found that interspecies differences, distortion of outcomes from laboratory environments and protocols, and poor methodological quality of many animal studies all contributed to low clinical utility.¹⁸ The overall success rate of drugs during clinical development stays low, and flawed preclinical research using animal models is widely considered a major reason for the gap between laboratory promise and clinical reality.¹⁹

Zebrafish and the Rise of Non-Mammalian Models

Not all animal testing involves mice, dogs, or primates. Zebrafish have become an increasingly popular model organism, especially for drug screening and toxicity testing. They are small, breed rapidly, develop quickly, have transparent embryos that let researchers watch organ development in real time, and share a surprisingly large portion of their genome with humans. These traits make it possible to screen thousands of compounds in living animals far more quickly and cheaply than mammalian studies allow.²⁰

Because zebrafish are whole organisms rather than cells in a dish, they capture the complexity of how a drug interacts with a living body, including how it is absorbed, distributed, and broken down. That gives them an advantage over cell-based high-throughput screening methods while still being fast enough for large-scale use.²¹ Zebrafish mutations can faithfully mimic many human disorders, and several compounds first identified through zebrafish screening have advanced to early clinical trials for conditions including a type of salivary gland cancer, a severe form of childhood epilepsy, and a rare blood disorder.²²

Organ Chips, AI, and the Technologies Edging In

The limitations of animal models have driven substantial investment in alternatives. Organ-on-a-chip devices, which are small microfluidic platforms lined with living human cells cultured under flowing fluid, can mimic the function of human organs with high accuracy. These chips have been used to model complex diseases and rare genetic disorders, to study how the body’s organs interact with each other, and to reproduce human responses to drugs, toxins, and infectious agents.²³ Multi-organ chip systems that connect models of the liver, kidney, gut, and other organs on a single platform are being developed to predict how drugs move through and affect the whole body, addressing a weakness of traditional cell culture studies.²⁴

Artificial intelligence is another rapidly growing tool. Machine learning models can now predict certain toxic properties of new compounds before any living thing is exposed to them, by analyzing patterns in chemical structures and large datasets of known drug-toxicity relationships.²⁵ Deep learning and multimodal data fusion are reshaping how toxicity is assessed, though these tools are still better at flagging likely problems than at replacing the full picture an animal study provides.²⁶ In practice, machine learning is increasingly used to select promising compounds early and to filter out those likely to fail for safety reasons, reducing the number of animals needed in later stages.²⁷

Regulators are taking notice. In 2025, the U.S. Food and Drug Administration announced a formal roadmap to reduce animal testing requirements in drug development, emphasizing human-centric approaches including stem-cell-derived organoids, organs-on-chips, computer modeling, and AI tools. The motivation is not just ethical; the FDA’s roadmap explicitly cites the poor predictive accuracy of animal models, high costs, and the clinical failure rates that result from interspecies biological differences.²⁸ Still, challenges remain for organ-chip technology, including scaling up manufacturing and achieving regulatory acceptance for new types of evidence.²⁹

The Regulatory History That Embedded Animals in Drug Testing

Animal testing became a regulatory requirement not through a gradual scientific consensus but through disaster. Two public health catastrophes shaped the modern system. In the 1930s, the sulfanilamide elixir disaster, in which a toxic solvent in an untested drug formulation killed over a hundred people, led to laws requiring safety testing before drugs could be sold. In the early 1960s, the thalidomide tragedy, in which a morning sickness drug caused severe birth defects in thousands of babies, prompted even more extensive animal testing requirements under the belief that thorough preclinical work would prevent similar catastrophes.³⁰ The irony is that thalidomide’s effects in humans were not reliably predicted by the animal testing protocols of the time, yet the regulatory response was to require more of it rather than to question the fundamental approach.

The ethical framework governing animal research has also evolved. The “3Rs” framework, first articulated in 1959, calls for the replacement of animals with non-animal methods where possible, the reduction of the number of animals used, and the refinement of procedures to minimize suffering. This framework is now incorporated into legislation worldwide and has pushed laboratories toward smaller study sizes, better anesthesia, and more humane endpoints.³¹

Why Lab Conditions Themselves Affect What Studies Find

A rarely discussed factor in the reliability of animal research is the environment in which the animals live. Standard laboratory housing for rodents has traditionally been austere: a plastic box, bedding, food, and water. Research shows that this barren environment raises stress hormones, increases anxiety-like behavior, and produces physiological changes that can skew experimental results in unpredictable ways. Providing enrichment like nesting material, shelters, and objects to explore lowers stress-related hormones, reduces elevated heart rate and blood pressure, and produces calmer, more behaviorally normal animals.³²

Some researchers have worried that enrichment would introduce variability and make results harder to reproduce. The evidence suggests the opposite. A broad study found that adding nesting material and shelters to mouse housing did not meaningfully change the parameters that researchers typically measure in biological and medical studies, and concluded that enrichment could be applied widely without skewing results or compromising comparisons to past data collected under barren conditions.³³ In fact, giving animals the chance to express normal species behavior likely produces a more accurate model of the biological processes researchers are trying to study, improving both the welfare of the animals and the validity of the science.³⁴ This means that some fraction of historically failed translations from animal to human may have been partly a product of stressed-out animals producing unrepresentative biology, rather than an inherent failure of the animal model itself.