FDA is deleting the word 'animal' from its safety-testing rules. What changes?

FDA has issued a direct final rule that replaces the words "animal test" and "animal study" with "nonclinical test" and "nonclinical study" throughout five parts of its drug and biologics regulations, effective 4 February 2027 [s1]. The change is purely terminological — FDA says it "adds no new requirements" — but it is not cosmetic: it removes language that treated animal testing as the default way to establish a drug's safety before human trials [s1].

What the rule actually does

The direct final rule amends 21 CFR parts 312, 314, 315, 361 and 601 — the sections governing investigational new drug applications, new drug and biologics approvals, and related safety reporting [s1]. It substitutes "nonclinical" for "animal" and for "preclinical," adds formal definitions of "nonclinical test" and "nonclinical study," and makes conforming edits so the regulations no longer imply that safety data must come from animals [s1]. FDA published it as a direct final rule — a route reserved for changes it considers noncontroversial — alongside an identical companion proposed rule [s1][s2]. If the agency receives significant adverse comments by 7 December 2026, it will withdraw the direct final rule and proceed through the ordinary proposed-rule process instead [s1][s2].

Nothing in the rule tells a drug developer to stop using animals or to adopt any particular alternative. It changes the vocabulary of the regulations so that non-animal methods sit on equal footing with animal studies as ways to satisfy the same underlying safety obligations [s1].

The statute behind it

The rule implements a change Congress already made. Section 3209 of the Food and Drug Omnibus Reform Act — enacted as part of the Consolidated Appropriations Act, 2023 (Public Law 117-328) — added section 505(z) to the Federal Food, Drug, and Cosmetic Act and broadened the definition of the nonclinical testing that can support an application [s1]. That statutory definition now lists, as acceptable methods, cell-based assays; organ chips and microphysiological systems; computer modelling; other nonhuman or human biology-based test methods; and animal tests [s1]. Animal testing remains on the list — it is one option among several, not the mandatory baseline it once was in the regulatory text [s1].

FDA's terminology rule brings the CFR into line with that statute. It is the housekeeping step that follows a policy already set in law.

The evidence case, and its limits

The argument for the shift is that some newer, human-based methods may predict human drug safety more accurately than animal models for certain questions, while using fewer animals. In April 2025 FDA published a roadmap to reduce animal testing in safety studies by replacing them, in a stepwise way, with what it calls scientifically valid new approach methodologies, beginning with the safety testing of monoclonal antibodies [s1]. The agency frames this as a way to "improve drug safety" and find more efficient methods, not merely to reduce animal use [s1].

The rule is careful not to oversell that case, and the honest reading tracks its caution. FDA states plainly that "there remain areas where animal testing is important and necessary" — for example, evaluating toxicities that arise through whole-body physiological interactions, such as the release of hormones, neurotransmitters and cytokines and the communication between organ systems, which an isolated assay cannot reproduce [s1]. It also notes that methods using human-derived cells may in other cases capture more relevant endpoints than an animal model [s1]. The upshot is not that one approach is superior across the board; it is that the right method depends on the question, and that the regulations should stop presuming the answer is always an animal.

The gating step is validation. FDA points to work by ICCVAM — the interagency committee that coordinates test-method evaluation — on building confidence in new approach methodologies through "flexible, fit-for-purpose validation strategies" that weigh a method's context of use and biological relevance rather than approving it in the abstract [s1]. In other words, a method does not become acceptable because the regulations now call it "nonclinical"; it becomes acceptable when it is shown to answer a specific safety question well. That is a narrower and more defensible claim than the way such changes are sometimes described, and it is why FDA is layering the transition through method-by-method guidance rather than a single switch. The broader scientific debate over how far these methods can substitute for animals is covered in new approach methodologies and animal testing; Europe's regulator is running a voluntary NAMs data-submission pilot to build the same kind of evidence base, and FDA has separately been rethinking the evidence it accepts for ultra-rare diseases.

What to watch

Whether FDA receives significant adverse comments by 7 December 2026 that force it to withdraw the direct final rule and re-propose it; whether the promised method-specific guidance — starting with monoclonal antibodies — actually validates alternatives for concrete safety questions; and whether "nonclinical" in the regulations comes to mean a real menu of methods or simply a renamed default.