Stability Testing and Shelf Life: How a Supplement’s Expiration Date Is Set

An expiration date looks like a fact printed on a bottle. It isn’t. It’s a claim: a promise that the product still contains what the label says, and is still safe and acceptable to use, on the day it’s printed for. Like every claim you put on a label, you have to be able to back it.

Most brands set that date the wrong way: they copy a competitor, round to a comfortable number, or let a broker pick one. Then a marketplace, a retailer, or a regulator asks how they arrived at it, and there’s no answer. Stability testing is how you get an answer: a documented study that measures whether your product actually holds up over the time you want to claim. Set the date from that data and it survives scrutiny; set it any other way and you’re printing a number you can’t defend.

Educational overview: not legal, regulatory, or medical advice. Requirements change and vary by jurisdiction and sales channel. Last reviewed July 2026.

Short answer. A supplement’s expiration date is a substantiated claim, not a default. As of mid-2026, FDA does not require a dietary supplement to carry a date at all, but if you print one, FDA’s stated position is that it should be supported by data. Stability testing supplies that data: you store the finished product under defined conditions and measure potency, safety, and appearance at set intervals. Real-time testing runs the full length of the claim; accelerated testing ages the product faster to support a provisional date while real-time confirms it.

Best for: Founders and brand owners setting a first expiration date, and existing brands re-checking one at a transfer or reformulation.

Key decision: Whether to launch on a provisional date backed by accelerated data (with real-time running in parallel), ship a manufacture date while real-time completes, or wait for full real-time data before you claim a shelf life.

Apollo path: Apollo coordinates stability studies through vetted third-party labs and formulates for the shelf life you need, then gives you the documented data package behind the date.

What an expiration date actually claims, and what the FDA does and doesn’t require

Start with the regulation, because most published guidance skips it and gets the burden backwards.

As of mid-2026, the FDA does not require a dietary supplement to carry an expiration date, a “best by” date, or any shelf-life date. The dietary supplement cGMP rule (21 CFR Part 111) does not mandate one. You can read the choice directly in the reserve-sample rule: § 111.83(b)(3) tells a manufacturer to hold reserve samples “for 1 year past the shelf life date (if shelf life dating is used), or for 2 years from the date of distribution” when it isn’t. The regulation is written around dating being optional. When the FDA issued Part 111, it said the same thing plainly in the rule’s preamble: “the final rule does not require that you establish an expiration date.”

So where does the burden come from? Two places, and they’re the part that matters.

First, you have to meet your own specifications. Part 111 requires that “for each dietary supplement that you manufacture you must establish product specifications for the identity, purity, strength, and composition of the finished batch” (§ 111.70(e)), and that you verify those specifications are met (§ 111.75). Strength is the one tied to time: if your label declares 500 mg of an active, the product has to contain what your specification says, and a degrading active is a moving target.

Second, any date you print has to be true. In that same rulemaking, FDA was direct: “any expiration date that you place on a product label should be supported by data.” The agency even declined to prescribe how. So the date is yours to make and yours to substantiate; print one you can’t defend and you’ve created risk (a potentially misbranded product, a marketplace dispute, a retailer chargeback) without making the product any safer.

That’s the honest frame the SERP’s stability guides leave out. Dating is optional; substantiation is not. Stability testing is how you generate the substantiation.

The manufacture-date alternative. Because a date isn’t required, you have an option that surprises a lot of founders: print a manufacture date instead of an expiration date. It’s a factual statement of when the batch was made: no long-term data needed, because it claims nothing about the future. Many brands launch this way while a real-time study runs, then add a substantiated “best by” date to later runs. It’s not a loophole; it’s a defensible way to reach market without printing a claim you haven’t earned. (Some marketplaces and retailers require a legible date of some kind; confirm your channel’s rules first.)

The two ways to prove a shelf life: real-time and accelerated

There are two studies behind almost every defensible date, and the difference between them is just time versus temperature.

Real-time (long-term) testing stores the finished product at conditions meant to stand in for a normal shelf, then measures it on a schedule across the full period you want to claim. The industry runs this at 25 °C and 60% relative humidity: the long-term condition drawn from the ICH stability guidelines that supplement makers borrow from pharmaceutical practice. (FDA doesn’t mandate a specific protocol for supplements, so the industry adopts ICH as the accepted standard.) Real-time data is the most defensible you can get, for the obvious reason: it’s the product actually aging. The catch is equally obvious. A 24-month claim isn’t fully confirmed until 24 months have passed.

Accelerated testing buys you speed. You store the product under deliberate stress (the standard accelerated condition is 40 °C and 75% relative humidity), so the chemistry that would take years on a shelf plays out in months. A typical accelerated study runs about six months, with measurements at 0, 1, 2, 3, and 6 months. A common industry rule of thumb, from reaction-rate models (the Arrhenius / Q10 approach), treats about six months of accelerated storage as a rough stand-in for roughly two years at room temperature. It’s a planning estimate, not a promise: heat doesn’t accelerate every reaction equally, and some ingredients fail in ways heat exaggerates or hides.

That limitation is why accelerated data doesn’t stand alone. The accepted practice, which FDA has allowed, is to use accelerated results to support a provisional expiration date so you can launch, while a concurrent real-time study runs to confirm or correct it. If real-time later shows the product fading faster than predicted, you shorten the date on future runs. You’re not guessing; you’re launching on a defensible projection and verifying it as it ages.

Two more conditions come up for specific products. Intermediate (30 °C / 65% RH) bridges the gap when accelerated data shows a borderline “significant change” result (a middle stress level that reads more realistically than 40 °C). Photostability exposes the product to defined light (the ICH light-exposure test) to see whether it degrades or discolors on a lit shelf or in a clear bottle; it matters most for light-sensitive actives and clear-container liquids.

Conditions and intervals reflect ICH-derived industry practice, which FDA does not mandate for supplements; your protocol may differ by product and channel.

What a stability study measures, and how the date gets read off the data

A stability study isn’t one test. It’s a schedule of pulls, and at each pull the lab measures several things at once. Skip an attribute and you’ve substantiated potency while staying blind to whether the product turned or separated. At each pull, a well-built study measures:

• Potency: the amount of each active, against your label claim. This is the attribute the date usually turns on.
• Microbial quality: total counts and the absence of specified organisms, so the product stays safe as it ages, not just strong. This matters more the more water a product contains.
• Organoleptic (sensory): appearance, color, odor, and taste. A supplement can hold potency and still fail because it darkened, clouded, or went off-flavor.
• Physical and packaging compatibility: pH, viscosity, and separation in liquids; hardness and moisture pickup in tablets and capsules; and whether the product and its container are quietly changing each other (a closure that lets oxygen in, a liquid that leaches from or sorbs into its bottle).

Two things make the data hold up. Studies are typically built on at least three production batches, so you’re characterizing the formula and process, not one lucky lot. And the product is stored in its final, market-ready container and closure (the same bottle, liner, and cap you’ll actually sell), because packaging is part of what determines shelf life, not a detail you can swap later.

Here’s the part every thin guide omits: how you actually read a date off the numbers. The date isn’t the end of the study; it’s the last pull point where the product still meets every acceptance limit in your specification, potency included. Because actives degrade, formulators build in formulation overage, so a fresh unit often starts above 100% of label claim and drifts down toward it. Watch a simplified, illustrative real-time table for a liquid:

Illustrative only. Real degradation curves depend on the ingredient, the formula, the packaging, and the conditions; these numbers are not a prediction for any product.

Read it the way a quality team does. The product launches at 115% of claim (that’s the overage, not a bonus dose) and fades. It still meets a full 100% of the label at month 18. By month 24 it has dipped below claim, so a “24-month” date would promise something the product no longer keeps. The defensible date is 18 months: the last point where potency, microbial, and sensory results all pass, not the number that looked nicest on the box. (You set the acceptance limit in your specification under § 111.70(e); most brands hold potency at or above 100% of label claim through the date.) The date is an output of the data, read against your spec. Nothing more.

Why liquids run a harder clock than capsules

If your product is a liquid (a shot, a tincture, a syrup, a ready-to-drink), its shelf life is a different, and usually shorter, problem than a dry format’s. This is where a liquid-first manufacturer earns its keep, because the reason is chemical, not incidental.

Water is what drives most degradation. In a dry capsule or powder, water activity is low and the actives mostly sit still. Dissolve those same actives in a liquid and you’ve put them in the medium where reactions happen: hydrolysis cleaves susceptible molecules, oxidation proceeds with dissolved oxygen, pH drifts as the matrix ages and can drag sensitive actives down with it, and microbial growth becomes a live risk that a dry powder mostly escapes. A liquid also spends its whole life in intimate contact with its packaging, so container compatibility (oxygen ingress through the closure, light through a clear bottle, leaching or sorption between product and plastic) moves from a footnote to a headline.

Three consequences follow, worth planning for before the bench work locks. Liquids typically need more overage: because actives fade faster in solution, the formula has to start further above label claim to still meet it at expiration; set that amount from stability data, not a guess (see formulation overage). Preservation and pH are part of the shelf life: the preservative system, pH, and antioxidant strategy are what buy the months you want to claim; get them wrong at the bench and no amount of testing rescues the date. Packaging is a stability variable: the bottle, liner, and closure change the number the study returns, which is why stability is run in the final container and why picking packaging late can force a re-test.

None of this makes liquids a worse format. It makes them one where formulation judgment and stability testing have to work together from the start: the pairing a liquid-first shop is built around. What drives the cost of a liquid supplement covers where that complexity shows up in a quote.

The change that re-opens your shelf-life claim: the one brands forget at transfer

Here’s the trap that catches established brands, and it’s the piece the whole stability-testing SERP leaves out. A shelf life you substantiated once is substantiated for that product, made that way, in that package, at that site. Change one of those, and you may have changed the thing your data described, which means the date may no longer be yours to claim.

The changes that can re-trigger stability work:

• The formula: a new active, ratio, or excipient, or even a different form of the same ingredient (a different salt, a different extract ratio) can degrade on its own curve.
• The ingredient source: a new supplier or grade can carry different impurities, moisture, or aging behavior.
• The packaging: a new bottle, liner, closure, or a glass-to-plastic switch changes the barrier around your product, one of the biggest levers on a liquid’s shelf life.
• The process: a different mixing method, fill temperature, or hold time can change how the product ages.
• The manufacturing site: moving production to a new facility is a process-and-site change, and this is the one brands forget.

That last one is where it bites. When a brand switches manufacturers (for better MOQs, faster reorders, or to escape a quality problem), the shelf life comes along in the assumptions but not always in the data. Different equipment, different components, sometimes an adjusted process: the new run won’t necessarily age like the old one. A careful transfer treats the existing shelf life as a claim to re-confirm, not inherit, usually putting the first runs from the new site on stability (accelerated to bridge, real-time to confirm) while you keep selling on the established date.

This is why stability and a manufacturer transfer are linked, and why the honest version of a transfer plans for it. The benchmark-to-pilot step of a transfer runs roughly six to eight weeks on Apollo’s floor (a benchmark sample, a spec match, and a pilot run), but stability re-confirmation, when a change triggers it, runs on its own method-inherent clock beyond that pilot, because you can’t accelerate real time. Building that into the plan up front is the difference between a clean switch and a surprise. The manufacturer transfer playbook walks the full sequence; the stability point is simply that a site change is a change, and changes get re-checked.

How long stability testing takes: the honest answer

“How long does stability testing take?” has a real answer, and it’s not a number a manufacturer can shorten by wanting to.

Real-time testing takes as long as the claim. A 24-month shelf life isn’t fully confirmed until 24 months of data exist. There’s no honest way around that; anyone claiming to generate two years of real-time data in less than two years is describing accelerated data, which is a projection, not the confirmation.

Accelerated testing takes about six months to produce the provisional date you launch on. That’s the practical long pole for a first launch: not the mixing, the fill, or the labeling, but the wait for enough aging data to stand behind a date.

Which leads to the one planning move that matters: start stability early and build the launch around it. Put representative material on stability as soon as the formula and packaging are locked, and decide up front which path you’re taking: a provisional date with concurrent real-time, a manufacture date while real-time runs, or waiting for full data. Formats and fills move on their own schedules (material lead alone can run around 30 days), but stability rewards being started first and punishes being remembered last.

What to have, and what to ask

Whether you’re setting a first date or checking one at a transfer, this is the documented package behind a defensible shelf life. Ask for it by name; a real manufacturer produces it as a matter of course.

What a defensible shelf-life claim is built on:

• A written stability protocol: conditions, pull schedule, and the analytical methods used at each pull.
• The storage conditions you’re testing against (real-time, accelerated, and photostability if the product warrants it).
• A defined acceptance limit for each attribute (potency against label claim, microbial limits, sensory and physical limits), the lines the product has to stay inside.
• The final, market-ready container and closure used in the study, not a stand-in.
• A change-control list naming the changes (formula, source, packaging, process, site) that re-trigger stability.
• The data package and CoAs (real numbers at each pull, from vetted third-party labs) that you can show a marketplace, a retailer, or a regulator.

Questions worth asking any manufacturer, Apollo included:

• Is my expiration date supported by real-time data, accelerated data, or a provisional accelerated date with real-time running concurrently?
• What acceptance limit sets my date, and does potency hold at or above label claim through it?
• Was the study run in my actual production packaging, on more than one batch?
• What overage is built in, and is it set from stability data or assumed?
• Which changes on your side would re-trigger stability, and does moving my product to your site count as one?
• Can I see the stability data behind the date, not just the date?

If those answers come back clearly and in writing, you’re looking at a substantiated shelf life. If the date arrives with no study behind it, that’s your signal to keep asking: the date is a claim, and a claim you can’t document is a claim you shouldn’t print.

About this information

This article is an educational overview for brand owners and operators. It is not legal, regulatory, or medical advice. Laws, regulations, and marketplace policies change frequently and vary by jurisdiction and by where and how a product is sold; the regulatory points here reflect our reading of the relevant sources as of July 2026 and may not be current or complete for your situation. Verify current requirements with qualified counsel or regulatory professionals before acting. Apollo Future Labs supports manufacturing execution: formulating for a target shelf life, coordinating stability studies through vetted third-party labs, and documenting the data behind a date; your counsel and quality team decide your compliance strategy and what you claim on a label.