Do I need a real-time stability study, or is accelerated data enough?

Read the first section before the rest. Almost everything published about stability testing is written for drug products, because that is where the federal requirement lives. Many of the people typing this question do not make drug products. Whether any of what follows is a requirement for you, or a benchmark you are choosing to meet, depends entirely on what your product is under United States federal law.

On this page: Which rule is talking to you · What the drug rule says about accelerated data · Where ICH Q1A fits, and where it does not · What FDA’s inspection record shows · If no federal rule requires it, what does · Why the accelerated number is the one that moves · Four things to check

First: which rule is talking to you

Drug products

Parts 210 and 211 contain the minimum current good manufacturing practice for the manufacture, processing, packing or holding of a drug (21 CFR 210.1(a)), and part 211 states its own scope as the minimum practice for preparation of drug products for administration to humans or animals, excluding positron emission tomography drugs and medical gases (211.1(a)). Inside it sits the requirement: to assure that a drug product meets applicable standards of identity, strength, quality and purity at the time of use, it shall bear an expiration date determined by appropriate stability testing described in 211.166 (211.137(a)). Expiration dates are related to any storage conditions stated on the labeling, as determined by those same studies (211.137(b)), and appear on labeling in the location 201.17 sets (211.137(d), 201.17).

The exceptions travel with the requirement, and there are five of them:

• Homeopathic drug products are exempt from the expiration-dating section (211.137(e)) and are held instead to a written assessment of stability based at least on testing or examination for compatibility of the ingredients, and on marketing experience indicating no degradation over the normal or expected period of use, evaluated in the same container-closure system in which the product is marketed (211.166(c), (c)(1), (c)(2)).
• Allergenic extracts labeled “No U.S. Standard of Potency” are exempt from both sections (211.137(f), 211.166(d)).
• New drug products for investigational use are exempt from the expiration-dating requirements, provided they meet appropriate standards or specifications as demonstrated by stability studies during their use in clinical investigations; and where such products are to be reconstituted at the time of dispensing, their labeling shall still bear expiration information for the reconstituted drug product (211.137(g)). Separately, an investigational drug for use in a phase 1 study remains subject to the statutory current good manufacturing practice requirement at 21 U.S.C. 351(a)(2)(B), and it is the production of such a drug that is exempt from compliance with part 211; that exemption stops applying once the investigational drug has been made available for use in a phase 2 or phase 3 study, or the drug has been lawfully marketed (210.2(c)).
• Human OTC drug products whose labeling bears no dosage limitations and that are stable for at least three years as supported by appropriate stability data: pending consideration of a proposed exemption published in the Federal Register of September 29, 1978, the expiration-dating requirements shall not be enforced for them (211.137(h)). Read that clause to its end. The condition is holding stability data showing three years. It relieves the printed date, not the data behind it, and it says nothing about 211.166.
• OTC drug products that are also foods. Pending the same proposed exemption, part 211 shall not be enforced for OTC drug products if the products and all their ingredients are ordinarily marketed and consumed as human foods; parts 110 and 117, and where applicable parts 113 through 129, are applied instead (211.1(c)).

One more that decides scope rather than substance: a person who engages in only some of the operations these parts cover need comply only with the regulations applicable to the operations they are engaged in (210.2(b)).

Dietary supplements

Part 111 applies if you manufacture, package, label or hold a dietary supplement, including one you manufacture that another person packages or labels, and one imported or offered for import; the holding requirements do not reach you if you are holding at a retail establishment for the sole purpose of direct retail sale to individual consumers, which does not include a warehouse or other storage facility (21 CFR 111.1(a), (a)(1), (a)(2), (b)).

Part 111 has no stability-testing section and no expiration-dating section. It names shelf life three times and conditionally every time: reserve samples are retained one year past the shelf life date “if shelf life dating is used”, or two years from the date of distribution of the last associated batch (111.83(b)(3), 111.465(b)), and written records are kept on the same either-or basis (111.605(a)). The clause assumes a shelf life date may or may not exist. It never asks you to create one.

What part 111 does require is that, for each dietary supplement you manufacture, you establish product specifications for the identity, purity, strength and composition of the finished batch, and for limits on contamination that may adulterate it (111.70(e)), and that you verify the finished batch meets them, for a subset identified through a sound statistical sampling plan or for every finished batch, by selecting established specifications that would verify the system is producing a supplement meeting all product specifications (111.75(c), (c)(1)). That provision carries its own exemption route: you may exempt a specification from verification where the selected tests cannot verify it and there is no scientifically valid method for testing it at the finished batch stage, if you document why other controls will ensure it is met, and your quality control personnel review and approve that documentation (111.75(d)(1), (d)(2)). All of that is a test on the finished batch. It is not a claim about month 24.

Part 111 also states that other applicable statutory provisions and regulations under the act still apply to you (111.5). One of them is the misbranding provision: a food is deemed misbranded if its labeling is false or misleading in any particular (21 U.S.C. 343, 343(a)). A dietary supplement is a food, and a printed date is a statement on the label. So the position is not that dates are unregulated. It is that no clause requires you to print one, and a different clause governs whether what you printed is true. Whether a particular date is true is a question about your product and your data. No rule settles it in advance.

Cosmetics

No expiration-dating or stability-testing requirement appears in 21 CFR part 700 or in the cosmetic provisions of the Federal Food, Drug, and Cosmetic Act as amended by the Modernization of Cosmetics Regulation Act of 2022. We searched both, as read at eCFR and the United States Code on August 18, 2026, against the terms “stability”, “shelf life” and “expiration” and found none. A period-after-opening symbol or a best-by date on a cosmetic is a commercial and, in some export markets, a foreign regulatory decision. It is not a United States federal requirement.

Conventional food

21 CFR part 117 carries no expiration-dating or stability requirement either, on the same reading. The federal exception in the sources read here is infant formula: in addition to the applicable labeling requirements of parts 101 and 105, the label shall bear a “Use by” date, the month and year selected by the manufacturer, packer or distributor on the basis of tests or other information showing that until that date, under the conditions of handling, storage, preparation and use prescribed by label directions, the formula will contain not less than the quantity of each nutrient set forth on its label when consumed, and will otherwise be of an acceptable quality (21 CFR 107.20, 107.20(c)). And the shelf life behind that date is tested rather than asserted. For a new infant formula the manufacturer collects, from each manufacturing site and at the final product stage, a representative sample of the first production aggregate of packaged, finished formula in each physical form, evaluates the levels of all nutrients required under 107.100 and all other nutrients it added, and repeats that testing every four months throughout the shelf life of the product (21 CFR 106.91(b), (b)(1)(i)), unless FDA exempts it on request on analytical data showing the stability of the new formula will likely not differ from formulas of similar composition, processing and packaging for which there are extensive stability data ((b)(1)(ii)). Each subsequent production aggregate gets the same evaluation, repeated at the end of the shelf life of the product ((b)(2)). That testing does not have to evaluate mineral levels ((b)(5)), and all quality control testing is conducted using appropriate, scientifically valid test methods ((c)). Where the results do not substantiate the shelf life, the manufacturer shall address, as appropriate, all production aggregates released and pending release that the results implicate, such as by conducting the testing on a subsequently produced aggregate to substantiate the shelf life or by revising the use-by date so that it is substantiated by the stability testing results ((b)(3)).

That is a complete federal shelf-life regime for a food, and it exists for one product class. If you do not make infant formula, none of it reaches you. What it is useful for is the standard it sets in the open: a date, the testing that substantiates it, and a written answer for the day the testing does not.

This list of product classes is not exhaustive. Medical devices, animal food, tobacco and biologics each run under their own parts, state law is not covered here, and a customer contract can require what a regulation does not.

What the drug rule actually says about accelerated data

21 CFR 211.166 has a chapeau and a list, and the list is part of the requirement. There shall be a written testing program designed to assess the stability characteristics of drug products; the results of that testing shall be used in determining appropriate storage conditions and expiration dates; and the written program shall be followed and shall include (211.166(a)):

• sample size and test intervals based on statistical criteria for each attribute examined, to assure valid estimates of stability ((a)(1));
• storage conditions for samples retained for testing ((a)(2));
• reliable, meaningful and specific test methods ((a)(3));
• testing of the drug product in the same container-closure system as that in which it is marketed ((a)(4));
• testing of drug products for reconstitution at the time of dispensing, as directed in the labeling, as well as after they are reconstituted ((a)(5)).

Then the paragraph that answers the question. 211.166(b) is three sentences, and each one does separate work (211.166(b)):

1. An adequate number of batches of each drug product shall be tested to determine an appropriate expiration date, and a record of such data shall be maintained.
2. Accelerated studies, combined with basic stability information on the components, drug products and container-closure system, may be used to support tentative expiration dates, provided full shelf life studies are not available and are being conducted.
3. Where data from accelerated studies are used to project a tentative expiration date beyond a date supported by actual shelf life studies, there must be stability studies conducted, including drug product testing at appropriate intervals, until the tentative expiration date is verified or the appropriate expiration date determined.

So for a drug product the answer is settled in the text, and it is not “accelerated or real-time”. Accelerated data is not an alternative to a real-time study. It is what you may lean on while one is running. The permission carries two conditions at once, joined by “and”: that full shelf life studies are not available, and that they are being conducted. Satisfying one without the other does not satisfy the clause. And the permission is temporary by its own terms. The tentative date stands until it is verified by real-time data or a different date is determined.

The word doing the most work is tentative. It is in the regulation, not in a guidance document, and it is the whole of the answer.

Where ICH Q1A fits, and where it does not

ICH Q1A(R2) is where the familiar numbers come from, and it states its own scope in its first paragraph: it defines the stability data package for a new drug substance or drug product that is sufficient for a registration application within the three regions of the EC, Japan and the United States, and it does not seek necessarily to cover the testing for registration in or export to other areas of the world. It also says alternative approaches can be used where there are scientifically justifiable reasons (ICH Q1A(R2), introduction). Its stated purpose is to provide evidence on how quality varies with time under temperature, humidity and light, so as to establish a re-test period or a shelf life and recommended storage conditions, and its conditions were chosen for climatic zones I and II (§1.3).

For the general case of a drug product it sets long-term storage at 25 °C / 60 % RH or 30 °C / 65 % RH with twelve months of data at submission, and the accelerated condition at 40 °C / 75 % RH with six months (§2.2.7.1). The intermediate condition of 30 °C / 65 % RH for six months is conditional rather than standing: where 30 °C / 65 % RH is itself the long-term condition there is no intermediate condition, and where the long term is run at 25 °C / 60 % RH it is significant change during the six months at the accelerated condition that calls for the intermediate testing (§2.2.7.1). Long-term testing should be continued for a period of time sufficient to cover the proposed shelf life (§2.2.7). Where the available long-term data do not cover the proposed shelf life at approval, the guideline describes a stability commitment to continue the studies afterwards; where three production batches already cover the proposed shelf life, it considers a commitment unnecessary (§2.2.8). Its companion, ICH Q1E, exists specifically to describe when and how extrapolation beyond the available long-term data can be considered (ICH Q1E §1.1).

Mark the standing before you borrow it. ICH Q1A(R2) and Q1E are harmonized guidelines, not United States law. They are written for a registration application for a new drug. Nothing in them makes them a requirement for a dietary supplement, a cosmetic or a conventional food. What they are is the benchmark the field already speaks: when a customer, a certification body or an auditor asks how your date was set, an ICH-shaped answer is understood without translation. Borrowing it is usually a good choice. It is not the same thing as being under it, and the difference matters when someone asks you which one you are claiming.

What FDA’s inspection record shows

The published inspection observation record is the one place you can see how this fails in practice rather than how it is supposed to work. Across the two clauses that govern stability testing and expiration dating for drug products, the record holds 2,044 observations.

FDA inspection observations, 21 CFR 211.166 (stability testing) and 211.137 (expiration dating), fiscal years 2009 to 2026. Counts are of clause citations in inspection observations.

Clause	What the clause requires	Observations
211.166(a)	A written stability testing program, followed, whose results are used to set storage conditions and expiration dates	1,107
211.166(a)(3)	Reliable, meaningful and specific test methods	224
211.166(b)	Enough batches tested to determine the date; accelerated data only for tentative dates	207
211.166(a)(1)	Sample size and test intervals based on statistical criteria	42
211.166(a)(2)	Storage conditions for samples retained for testing	41
211.166(a)(4)	Testing in the same container-closure system as marketed	40
211.166(c)(1), (c)(2)	The homeopathic written assessment of stability, in the marketed container-closure system	40
211.166(a)(5)	Testing of products for reconstitution	4
211.137(a)	An expiration date determined by appropriate stability data	312
211.137(d)	The date appearing on labeling as prescribed	16
211.137(b)	The date related to the labeled storage conditions	11
Total		2,044

Inside the 207 under 211.166(b), 45 name accelerated data specifically. Thirty-two say that accelerated studies combined with basic stability information, used to support tentative expiration dates, were not supported with ongoing full shelf life studies. Thirteen say that where accelerated data projected a tentative date beyond what actual shelf life studies supported, the stability studies and the testing at appropriate intervals needed to verify it were not conducted. Those 45 span fiscal years 2009 through 2026, and every one of them sits in the Drugs program area.

What these figures do not evidence. Of the 1,107 observations under 211.166(a), 694 say there was no written stability testing program at all and a further 305 say the written program was not followed. A clause cited because a program was absent is evidence that programs go missing. It is not evidence that the programs which do exist are unsound, and once both are a number in a table they read identically. These are counts of clause citations in inspection observations, one row per citation, not counts of firms, of inspections, of warning letters or of enforcement actions, and a single inspection can contribute several rows. Every observation counted here comes from a drug-program inspection, so none of it says anything about a supplement, cosmetic or food maker, for the plain reason that the clause it was written under does not reach them. Fiscal year 2026 was still open at the read date, so its figure is partial.

Counted from FDA’s published inspection observation records, read August 18, 2026. The clause selection and the grouping are ours.

If no federal rule requires it, what does

Two things usually do, and neither is a regulation.

A certification scheme, where you hold one. NSF/ANSI 455-2, the dietary supplement GMP standard, requires that for all products bearing an expiration date or a statement of product shelf life, the shelf life be supported by data, and cites the preamble to the part 111 final rule as its basis (NSF/ANSI 455-2 §4.6.21). Its supporting provisions are written as recommendations rather than requirements, and the difference is visible in the verbs: a stability program using scientifically valid methods should be in effect, with ICH Q1 given as an example (§4.6.21.1); procedures should be established, and the program may include long-term, intermediate and accelerated studies as appropriate (§4.6.21.2); accelerated studies may be used to support initial shelf life claims and to evaluate changes such as proposed new primary packaging, and accelerated results should be verified with real-time studies (§4.6.21.3). Note the shape of that last line. It is the same answer the drug regulation gives, arrived at through a voluntary scheme, with “should” where the regulation says “must”. Its standing is what it is: a certification standard, which binds you where you hold or are seeking that certification and nowhere else.

A contract. Retailer and marketplace agreements, private-label specifications and customer quality agreements routinely ask for the data behind the date, and they are free to ask for more than any rule does. That is a commercial obligation rather than a regulatory one, and in our experience it is usually the one that arrives first.

Why the accelerated number is the one that moves

There is a pattern under this that is not specific to stability at all. Performance claimed from a test, a model or a published specification is generally worse when it is measured again in ordinary use, and the further the measurement moves from the conditions the publisher controlled, the worse the figure gets. That holds of averages rather than of any one case: where the published figure is a promise about a single instance, ordinary use can beat it as readily as fall short of it.

An accelerated study is exactly that shape. Six months at 40 °C and 75 % RH is a controlled condition chosen to make degradation happen fast enough to watch. The number that comes out of it is a projection about a product sitting in a warehouse, a truck and a bathroom cabinet, under conditions nobody controls. It can be right. It is still a proxy, and the drug regulation treats it as one by calling the date it supports tentative.

The other half is direction. Accelerated conditions do not speed every failure mode up by the same factor, and some they do not speed up at all: a slow physical change, a microbial count, a package that only loses its seal over months, an oxidation pathway that heat drives differently than time does. A product can come through six months at 40 °C looking clean and then fail at month eighteen at room temperature on the mode the heat never touched. That is not an argument against running accelerated studies. It is the argument for knowing which attribute your product actually fails on before you decide what the accelerated result means.

Four things to check before you spend

Take dated copies before you change anything. A stability protocol and an expiration-dating justification are controlled records, and revising one is your own quality unit’s decision rather than a tidy-up.

1. Settle your product class first. Everything above turns on it. The answer for a drug product and the answer for a dietary supplement are not the same answer, and a product that is both is governed as both.
2. Open your protocol and find the package. For a drug product, 211.166(a)(4) asks for testing in the same container-closure system as the one marketed. Where that clause does not reach you, the same question still decides whether your data means anything about the product you actually ship.
3. Find the sentence that says the real-time study is running. If your date rests on accelerated data, the drug rule’s permission is conditional on full shelf life studies being both unavailable and under way. What answers that is a protocol, batches on test and a pull schedule, not an intention to start one.
4. Decide now what happens when a pull point misses. Who is told, what the date becomes, what happens to product already made and already shipped, and whether the failing attribute changes the specification or the package. That is a written procedure or it is a scramble, and the difference is decided before the result comes in.

Related reading: how the study itself gets designed, how many batches, which conditions and how many time points. And where a customer rather than a rule is the one asking for the data, which document actually binds you, the quality agreement or the supply contract you already signed.