For more than a century, drug development has relied on animal studies as a gatekeeper for safety and toxicity. But in 2025, does that scientific justification still hold water? The original rationale — that animals are a safe proxy for humans — is deeply flawed, and institutional inertia (especially regulatory mandates) is the main reason it persists. It’s time to rethink animal testing in the context of modern, human-relevant methods.
The central flaw in the support for animal experiments is that these studies do not reliably predict human drug responses. Numerous reviews show that many compounds deemed “safe” in animals later cause toxicity in humans, and many drugs that fail in animals would have worked in people. A meta-analysis of over 3,000 drugs found that the absence of toxicity in animals offers “virtually no evidential weight” that adverse reactions will not occur in humans. Another systematic study showed animal toxicity had only moderate predictive accuracy (positive predictive value ~ 0.65, negative ~ 0.50) for human outcomes. Across fields like cancer, neurology, and inflammation, translation failure is rampant. The failure rate (over 90%) of new drugs in clinical trials underscores how poorly animal preclinical results translate.
Advocates of animal testing often claim that unknown toxicities must first be “caught” in animals. Most animal testing experiments are designed for extreme use cases to address rare safety risks, which, in practice, can be mitigated via design, dosing, monitoring, labeling, patient selection, pharmacovigilance, and controlled first-in-human protocols. Human genomic screening, biomarkers, microdosing, imaging, and adaptive trial monitoring now allow earlier detection of adverse responses. Regulations already permit updates to safety labeling or withdrawal if problems emerge post-approval, so the idea that animal testing is the only bulwark is outdated.
But there is a disconnect between announcements and reality. One of the vicious cycles here is regulatory lock-in. Because the U.S. FDA (and other regulators globally) continue to insist on animal test reports, drug developers have little incentive to invest in non-animal methods. That long-term neglect has slowed the validation, adoption, and scaling of alternative (“New Approach Methodologies,” or NAMs) such as organ-on-chip systems, microphysiological models, in vitro human cell models, organoids, and in silico toxicology models. The regulatory demand itself freezes the field in the past.
To its credit, the FDA has recently announced plans to phase out certain animal testing requirements, starting with monoclonal antibodies, replacing them with NAMs, AI, organoids, and computational toxicity models. Its “Roadmap to Reducing Animal Testing in Preclinical Safety Studies” lays out how validated human-relevant models could gradually supplant many categories of animal studies. Also, the 2022 FDA Modernization Act 2.0 removed the statutory requirement for animal testing in many cases. But despite this public stance, the FDA still frequently insists on animal data for INDs and new biologic approvals — creating a credibility gap between high rhetoric (e.g., support of the 3Rs) and regulatory practice.
Critics say: “We still need animals because there are few mature non-animal models.” But that is a self-fulfilling prophecy: the lack of regulatory acceptance disincentivizes investment, slowing progress. If the FDA truly discouraged obligatory animal tox studies and explicitly accepted NAM-based dossiers, the industry would rapidly pivot to develop, validate, and commercialize human-relevant toxicity assays. Market forces would drive innovation in in silico modeling, high-throughput human cell assays, multi-organ chips, quantitative in vitro-to-in vivo extrapolation (QIVIVE), and AI integrative prediction. Indeed, recent preprints and research show promising methods linking in vitro data to predicted in vivo toxicity. As confidence in human-centric models grows, reliance on animals would naturally fade, not because they were “eliminated” by fiat, but because they became scientifically inferior, more costly, and less predictive.
Given the weight of evidence, the continued mandatory use of animal studies for new drug safety is scientifically indefensible. Animal tests do not reliably forecast human risks; many adverse events can be managed by human-centric strategies; and regulatory fixation on animal models actively suppresses innovation in better alternatives. The real barrier is not science — it is regulatory inertia. If the FDA were compelled (via statute or guidance) to accept alternate data to support safety and validated non-animal data in lieu of animal tests, when alternate methods are insufficient on their own, the pharmaceutical industry would quickly follow. A true commitment to the 3Rs (reduce, refine, replace) requires dismantling the regulatory dogma that animal testing is indispensable. The future of safer, faster, more human-relevant drug development demands it.
