Third-Party Testing: What NSF and USP Actually Certify

What each mark actually certifies

The three programs a US buyer will encounter are not interchangeable. They differ in what they test, how often, and who they are designed for.

Program	What it verifies	Testing frequency	Built for
NSF (NSF/ANSI 173)	Label accuracy, toxicology review of added ingredients, no undeclared ingredients or unacceptable contaminants	Ongoing certification with audits	General buyers
NSF Certified for Sport	All of the above, plus screening for 280 substances banned by major athletic bodies	Ongoing, with banned-substance screening	Drug-tested athletes
USP Verified	Quality, purity, potency, performance and consistency against USP standards, plus FDA current good manufacturing practices	Ongoing audit and testing	General buyers
Informed Sport	More than 285 banned substances and impurities, ISO 17025 accredited methods	Every batch, before market release	Elite and tested athletes
Informed Choice	Same substance list as Informed Sport	At least 12 random lots per year	Active buyers, not drug-tested

Two distinctions are worth holding onto. First, Informed Sport tests every batch before release; Informed Choice tests random lots — the certifier’s own description is that Informed Sport means “every batch/lot of a product is tested prior to being released to the market,” while Informed Choice uses “regular blind testing” with “at least twelve lots/batches of the product… tested at random every year.”³ For a tested athlete the difference is the whole point. Second, NSF’s baseline standard and its Certified for Sport program are different marks; only the second screens for banned substances.¹

What none of them certifies

Efficacy. A certified bottle of a compound with no human evidence is a certified bottle of a compound with no human evidence. USP verifies quality, purity, potency, performance and consistency against its own standards — potency here means the product contains the declared amount, not that the declared amount does anything.² NSF confirms “what’s on the label is in the bottle.”¹ Neither organization is in the business of evaluating cognitive outcomes, and neither claims to be.

This matters more in this category than in most. The doses on the label of a brain supplement are frequently far below the doses used in the trials the marketing gestures at, and certification confirms the low dose rather than fixing it. A product containing 50 mg of bacopa monnieri when the meta-analyzed dose is 300–450 mg a day of a standardized extract⁶ can be certified, accurately labeled, contaminant-free, and useless in exactly the way the trial evidence predicts.

Certification also says nothing about the population an effect was found in. Alpha-GPC’s positive trials were in dementia patients at 1,200 mg a day, mostly alongside donepezil;⁷ a certified 300 mg alpha-GPC capsule aimed at a healthy 35-year-old inherits none of that evidence. No certification substitutes for reading the trial.

The problem the marks exist to solve

US supplements are not reviewed before sale. The NIH Office of Dietary Supplements puts it this way: “Medicines must be approved by FDA before they can be sold or marketed. Supplements do not require this approval,” and as long as a product contains no new dietary ingredient, “the company does not have to provide this safety evidence to FDA before the product is marketed.”⁴

There is a manufacturing rule — 21 CFR Part 111, “Current Good Manufacturing Practice in Manufacturing, Packaging, Labeling, or Holding Operations for Dietary Supplements” — and it does require a manufacturer to run at least one test to verify the identity of each incoming dietary ingredient.⁵ But CGMP is a set of obligations enforced by inspection after the fact, not a certificate issued after somebody tested the finished bottle. Compliance is assumed until an inspection or a complaint suggests otherwise.

The gap is not hypothetical in this category specifically. A 2020 analysis in Neurology: Clinical Practice tested US-sold cognitive-enhancement supplements and detected five unapproved drugs — omberacetam, aniracetam, phenibut, vinpocetine and picamilon — frequently at doses above pharmacological norms and often undeclared or mis-declared.⁸ FDA has separately stated that vinpocetine and phenibut do not meet the statutory definition of a dietary ingredient and has issued warning letters to nootropic sellers on that basis.⁹¹⁰¹¹ That is the failure mode third-party testing is designed to catch, and it is a real one.

How to check a certification in two minutes

1. Look the product up in the certifier’s own database, not on the seller’s page. NSF publishes a public searchable listing of certified products at listings.nsf.org, organized by category.¹² A logo on a sales page is a picture; a database entry is a record.
2. Check that the certified product is the one you are buying. Certification attaches to a specific product and often a specific formulation. A brand certifying one SKU does not certify its whole range.
3. For batch programs, check the lot. Informed Sport’s guarantee is batch-level; a bottle from an untested lot of a certified product is not covered.
4. Distinguish facility claims from product claims. “Made in a GMP-certified facility” and “made in an FDA-registered facility” are statements about a building. Registration is a filing requirement, not an endorsement, and neither phrase means the finished product was tested.
5. Treat an unnamed “third-party lab” as no certification at all. If the lab is not named and the report is not published with a lot number and a date, there is nothing to check.

What a good one looks like — criteria, not brands

Certification is one filter among several, and it is not the first one. In order:

• Per-ingredient doses printed on the panel. A certified proprietary blend is a verified mystery — the mark confirms the total weight is what the label says, which is exactly the number that was never the problem. The blend arithmetic is here.
• At least one ingredient at a dose that appears in a published human trial, in a population like yours.
• No red-list ingredient, certified or not. Certification of a product containing huperzine A does not make huperzine A appropriate to stack with a prescribed cholinesterase inhibitor.
• Then a current certification, looked up in the certifier’s database, for the specific product.
• A named legal entity and physical address, which every certifier requires anyway and most VSL-style sellers do not publish.

Note what this ordering implies. A certified product built on a compound with no evidence ranks below an uncertified single ingredient at a trial dose from a manufacturer with an address. Certification tells you the label is honest. The tier list tells you whether an honest label is worth reading.

Whether the premium is worth it

Certification costs the manufacturer money — testing, audits, annual fees — and that cost reaches the buyer. Whether the premium is rational depends entirely on which risk you are buying down.

Worth it, clearly: if you are drug-tested, if you are buying a botanical extract whose species is easy to substitute, or if you are pregnant, breastfeeding or managing a condition where a contaminant matters more than usual. Botanical identity is the classic case — several plant names in this category are shared between species, and “brahmi” refers to both bacopa monnieri and gotu kola, which are different plants with different evidence and different safety profiles.

Rarely worth a large premium: on simple, well-characterized, cheap single compounds where adulteration is commercially pointless — creatine monohydrate and L-theanine are the obvious examples. There is no economic incentive to adulterate an ingredient that already costs almost nothing.

Never worth it: as a reason to buy a compound you would otherwise have skipped. “It’s third-party tested” is the most common way a reader talks themselves into a product whose evidence they have not checked. The certification is real; the inference is not.

What would change our mind

A certification program that evaluated dose adequacy against published trials — verifying not just that the label is accurate but that the amount matches what was actually studied, in the population studied — would change the whole calculation, and we would recommend it above every criterion on this page. Nothing like it exists. Short of that, published assay data showing systematically better label accuracy in certified versus uncertified brain supplements specifically, rather than in supplements generally, would let us put a number on the premium instead of an argument.

Frequently asked questions

Is NSF or USP better?

They test overlapping things and neither is a general upgrade on the other. NSF’s baseline standard focuses on label accuracy, toxicology review of added ingredients and contaminants; USP verifies against its own quality, purity, potency and consistency standards plus FDA current good manufacturing practices. If you are drug-tested, the distinction that matters is not NSF versus USP but whether the specific mark screens for banned substances — NSF Certified for Sport and Informed Sport do.

Does “FDA registered facility” mean anything?

It means the facility filed a registration. It is not an approval, an inspection result, or a statement about the product, and FDA does not endorse products. Treat it as a phrase that appears on labels rather than as evidence.

Can a certified supplement still be underdosed?

Yes, and this is the single most useful thing to understand about certification. The mark confirms the bottle contains what the label declares. If the label declares 50 mg of an ingredient whose trials used 300 mg, certification confirms the 50 mg.

Why don’t the popular brain supplements carry these marks?

Certification requires disclosing the formulation to the certifier, submitting to audits, and paying for it — and the products that lean hardest on proprietary blends and origin stories have the least to gain from any of it. Absence of a mark is not proof of a bad product. In a category where FDA has issued warning letters over undeclared drugs, it is a reason to ask more questions rather than fewer.

Is ConsumerLab the same thing?

It is one of the three independent organizations NIH’s consumer fact sheet names alongside NSF and USP as offering quality testing. As with the others, its seal speaks to manufacturing and content, and the NIH’s caveat applies to all of them equally: these seals do not guarantee that a product is safe or effective.

Related reading

• Proprietary blends explained — why a certified blend is still unevaluable
• Nine ingredients that signal a bad brain supplement — the names certification will not save
• Do nootropics actually work? — the tier list, with the population behind every effect
• Racetams, noopept and phenibut — the drugs found in tested products
• How we rate evidence — what we accept as a source