FDA's September 22, 2026 direct final rule replaces 'animal test' and 'animal study' with 'nonclinical test' and 'nonclinical study' throughout its safety and reporting regulations, breaking the one-to-one equivalence that supplier attestations quietly relied on. Any wholesale contract or questionnaire that still treats 'nonclinical data on file' as a proxy for animal-study data no longer says what its drafters intended, and needs rewriting to force method-level disclosure.
On September 22, 2026, FDA published a direct final rule striking "animal test" and "animal study" from its safety and reporting regulations and substituting "nonclinical test" and "nonclinical study." FDA characterizes the change as terminology-only; the mechanics are detailed below.1
The consequence for a wholesale buyer is contractual rather than scientific. Every supplier attestation, resale agreement, and qualification questionnaire that treats "nonclinical data on file" as shorthand for animal-study data now points at a term covering at least three distinct method classes. The phrase has not become false. It has become non-informative, and a buyer relying on it as a proxy for animal-study data has no textual basis to challenge a supplier that satisfies it with a computer model.1,9
The September 22, 2026 direct final rule, sentence by sentence
It amends FDA regulations where existing text names animal testing as the method for generating safety information, and replaces those references with the nonclinical formulation. FDA's stated purpose is to remove language suggesting that animal testing is the only acceptable method, not to displace it.1
- Replaces terms such as "animal test" and "animal study" with "nonclinical test" and "nonclinical study" across FDA's safety and reporting regulations.
- Adds regulatory definitions of "nonclinical test" and "nonclinical study."
- Does not eliminate or prohibit animal studies.
- Does not change the evidentiary standards a submission must meet.
- States that it imposes no new costs or requirements on drug developers.
- Was issued alongside a companion proposed rule and an open public comment period.1
Read as a compliance event, the last three bullets are the ones that matter to a purchasing file. Because FDA disclaims any new requirement, no supplier acquires a disclosure obligation from the rule itself. The obligation has to come from the purchase contract, and the contract is the document that just stopped saying what its drafters thought it said.1
FDA already used 'nonclinical' as an umbrella term before this rule
The most common misreading of this action is that FDA introduced a new, looser term in September 2026. It did not. FDA's consumer update titled "Animal Welfare, Testing and Research of FDA-Regulated Products" already defined nonclinical testing as an umbrella covering three method classes, publishing that framework well before the September 2026 rule existed.9
- In vitro testing — petri dish or test tube work using human or animal cells and tissues.
- In silico testing — computer modeling.
- In vivo testing — animal studies.9
So the ambiguity is not new. What is new is where it lives. Until September 22, 2026, the umbrella definition sat in FDA explanatory material while the regulatory text that questionnaires and resale contracts quote verbatim still said "animal." The rule exported an internal FDA definition into the operative wording. Templates built on 21 CFR language are now citing a term the agency never restricted to animal data in the first place, and they inherited that breadth without anyone at the buyer's end approving it.1,9
The statutory hook: FDORA section 3209 and the FD&C Act 505(z) definition
The definitions in the rule implement section 3209 of the Food and Drug Omnibus Reform Act of 2022. That provision amended section 505 of the Federal Food, Drug, and Cosmetic Act and section 351 of the Public Health Service Act to clarify that non-animal test data may be sufficient to support an investigational new drug application or a biologics license application, and it added a definition of "nonclinical test" at FD&C Act section 505(z).1
That dating matters for contract archaeology. A supply agreement or qualification packet drafted before December 2022 that incorporates section 505 language about animal data by reference is incorporating superseded text. The 2026 rule did not create that gap; it closed the last place a reader could still find the old wording and assume it controlled.1
What a supplier's 'nonclinical data on file' attestation now fails to disclose
The phrase appears in three places in most wholesale paperwork: the supplier questionnaire, the representations section of the supply agreement, and downstream resale documentation passed to a reseller's own customers. In all three, it was drafted as a proxy. Nobody wrote it intending to accept a computer model, but nobody wrote it to exclude one either.1,9
| Attestation phrase | What the template assumed | What the phrase covers after September 22, 2026 |
|---|---|---|
| "Nonclinical data on file" | Animal study data | In vitro, in silico, or in vivo data, undifferentiated |
| "Nonclinical study completed" | A completed in-vivo protocol record | Any of the three method classes, no method named |
| "Meets applicable nonclinical requirements" | Animal-study-based requirements in FDA regulations | Requirements FDA has expressly declined to raise or lower by this rule |
| "Animal study data available on request" | Same as above, narrower wording | Still specific, but now out of step with the amended regulatory text it cites |
The fourth row is the one buyers overlook. A clause that still says "animal" remains specific and remains enforceable, but it now quotes wording FDA has removed from the cited provision. That mismatch is a citation problem, not a substance problem: the clause references section 505 language the amended CFR no longer contains. Specificity survives; the citation does not.1
The rule is a direct final rule with an open comment window, not settled text
A direct final rule is FDA's mechanism for changes it expects to be uncontroversial. It takes effect on the stated timetable unless significant adverse comment arrives, in which case the agency can withdraw it and fall back on the companion proposed rule issued the same day. Both the companion proposal and the comment period are part of this action.1
For a purchasing team, that changes the drafting method rather than the deadline. A questionnaire that hard-codes the amended regulatory phrasing this quarter is quoting text that can still be modified or withdrawn. The durable approach is to define the required disclosure inside the contract — naming the method class as a data field — and treat the CFR citation as a reference, not as the substance of the obligation.1
The rule sits inside a larger substitution program, not an isolated edit
The terminology change is one component of a multi-year FDA program. The agency's April 2025 "Roadmap to Reducing Animal Testing in Preclinical Safety Studies" and the FDA Modernization Act 2.0, which authorizes certain alternatives to animal testing for exemptions relating to investigating drug safety and effectiveness, are the policy spine the September rule bolts onto.10
CDER maintains published material on streamlined nonclinical studies and acceptable new approach methodologies, and FDA runs a standing NAMs program page covering the same substitution question across centers. On March 11, 2026, the agency issued draft guidance titled "General Considerations for the Use of New Approach Methodologies," six months before the terminology rule and pointing the same direction.3,4,5
FDA's press release attributes to Acting Commissioner Kyle Diamantas language framing the rule as part of an administration effort to expand accepted alternatives to animal studies, tied to HHS's Operation TrialBlazer initiative. The supply-side implication is direct: contract laboratories and upstream manufacturers have an expanding menu of accepted methods and, under this rule, no affirmative duty to tell a customer which one they used.1
What a re-qualification questionnaire needs to ask instead of 'is this nonclinical data'
The fix is to convert an attestation into a data field. An attestation asks whether a category was satisfied; a data field asks which member of the category was used. Only the second survives a term whose scope FDA has now formalized at three method classes.1,9
- Method class, named per study: in vitro, in silico, or in vivo. Required field, no free text, no "nonclinical" as an acceptable entry.
- For in vitro entries: cell or tissue system used and whether human- or animal-derived.
- For in silico entries: model identity, version, and the date of the run.
- For in vivo entries: species and the facility identifier where the work was performed.
- Study date, so the buyer can tell whether the record predates or postdates the December 2022 statutory amendment.
- The regulatory provision the supplier believes the study satisfies, cited with the version date of the text relied on.
- A change-notification clause obligating the supplier to notify the buyer in writing before substituting a method class on a study it has already certified.1,9
Item seven closes the gap the rule opens. FDA states the action imposes no new requirements on developers, which means a supplier that moves from an in-vivo method to an in-silico one breaches nothing under the amended regulations. It breaches the supply agreement only if the agreement says so. Written notice before substitution is the clause that makes the difference, and it costs nothing to add at renewal.1
For resellers passing documentation downstream, the same discipline applies one link further out. Research-use-only documentation that repeats a supplier's undifferentiated "nonclinical" language inherits the ambiguity wholesale and hands it to a customer who has even less ability to resolve it. Method-level identification in supplier qualification records is what keeps the chain legible when a buyer asks, eighteen months from now, what was actually run.1,9