You can evaluate most supplement claims with a short set of questions: who is legally allowed to say what, what kind of study sits behind the claim, whether it was in humans and large enough to mean anything, and whether anyone independent has reviewed the whole body of evidence. This guide teaches those skills, then condenses them into a checklist for any label, ad, or headline.

Key takeaways

  • The FDA does not approve supplements or their claims before sale; the company is responsible for having evidence, and the FTC acts only after deceptive ads run.
  • “Supports immune health” is a structure/function claim that requires no FDA review. A claim to treat, cure, or prevent a disease would make the product a drug.
  • Test-tube and animal studies cannot show that a supplement works in people. Human randomized controlled trials can; independent systematic reviews are stronger still.
  • “Clinically proven” is a legal claim, and FTC cases show how often it is used without a trial behind it.
  • PubMed, ClinicalTrials.gov, Cochrane, and the NIH fact sheets let you check a claim yourself in minutes.

Skill 1: Know who is allowed to say what

The first thing to understand about supplement claims is that nobody checked them before they reached you. The FDA states that it “does NOT have the authority to approve dietary supplements for safety and effectiveness” before they are sold, and it does not approve labels in advance either. Companies are responsible for ensuring their products meet the law; the agency’s role is after the fact, through inspections, label and website reviews, and adverse event monitoring (FDA 101: Dietary Supplements). Within that framework, labels can carry three kinds of claims, and the differences matter more than most shoppers realize.

Structure/function claims

These describe how an ingredient affects the normal structure or function of the body: “supports joint health,” “promotes restful sleep.” Under 21 CFR 101.93, a company must notify the FDA within 30 days of first marketing a product with such a claim and certify it has substantiation, but the FDA does not evaluate the claim. The rule also requires the familiar disclaimer: “This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.” When you see that asterisk, you are looking at a claim no regulator has reviewed.

Disease claims

The same regulation defines a disease claim as any statement, explicit or implied, that a product affects a specific disease or its symptoms, substitutes for a therapy, or uses its name, images, or citations to suggest a disease effect. A product bearing a disease claim is regulated as a drug. This is why careful marketers write “supports healthy blood sugar levels” rather than “lowers blood sugar in diabetes,” and why exact label wording tells you what a company could and could not defend.

Health claims and qualified health claims

Authorized health claims link a substance to reduced disease risk and must meet the FDA’s “significant scientific agreement” standard. Qualified health claims are weaker: the FDA explains they rest on evidence that “do[es] not meet the more rigorous ‘significant scientific agreement’ standard” and must carry qualifying language so consumers understand the limits. The FDA does not approve these; it issues a Letter of Enforcement Discretion stating it “will not object” to specific wording (FDA). Qualifying language such as “limited and not conclusive” is a built-in evidence rating; read it rather than skipping to the hopeful part.

Skill 2: Translate the marketing language

The Federal Trade Commission, not the FDA, polices advertising, and its standard is the one that matters for words like “clinically proven.” The FTC’s Health Products Compliance Guidance (December 2022) says health claims must be backed by “competent and reliable scientific evidence,” which for efficacy claims generally means randomized, controlled human clinical testing. It is explicit about what does not count: animal and in vitro studies “aren’t sufficient to substantiate health-related claims” without confirmation in humans, and consumer testimonials “are never sufficient to substantiate claims about the effects of a health product.” In one FTC example, a product “clinically proven to improve memory” relied on a study that measured only working memory, which did not support the broader claim.

Enforcement history shows how loosely these words get used. In 2015, marketers of Procera AVH gave up $1.4 million to settle FTC charges over claims that the supplement was “clinically proven to significantly improve memory, mood, and other cognitive functions” and would restore “10 to 15 years of memory loss” (FTC, 2015). In 2017, the FTC and the Maine Attorney General obtained a $6.57 million judgment against sellers of CogniPrin and FlexiPrin, whose ads claimed to reverse mental decline “by 12 years,” improve memory “by 44 percent,” and reduce the need for pain medication in “80 percent of users” (FTC, 2017). Precise-sounding percentages are not evidence; they are a reason to ask where the number came from.

A few translations help when reading any ad or label. “Clinically studied” means a study exists, not what it found, in whom, or at what dose. “Research shows” with no citation is an invitation to ask for one. “Doctor formulated” and “patented” describe origin, not effect. And “natural” says nothing about safety: NCCIH notes that kava has been associated with severe liver damage, that the FDA banned ephedra in 2004 after reports of heart problems and deaths, and that mercury, arsenic, and ricin are all natural (NCCIH).

Woman in a white coat reading at a computer in a pharmacy
Woman in a white coat reading at a computer in a pharmacy Photo: Loveleen Cherub / Pexels

Skill 3: Identify what kind of evidence you are looking at

Not all studies answer the same question. A finding that an extract kills cancer cells in a dish says nothing about whether swallowing a capsule does anything in a person, because it skips digestion, absorption, dose, and metabolism. The table below lays out the main study types in rough order of how much weight each can carry for a claim that a supplement helps people.

Types of evidence behind supplement claims, roughly ordered from weakest to strongest for questions about effectiveness in people
Evidence type What it can show What it cannot show Typical red flag
Test-tube (in vitro) study A compound has a biological activity on cells Whether a swallowed dose reaches tissues or changes anything in a person “Shown to kill cancer cells” on a product sold for general health
Animal study Effects and mechanisms in a living organism Whether the same happens in humans at a realistic dose Headline omits that the subjects were mice
Case report or testimonial That an outcome occurred in one person Whether the supplement caused it “Real customer results” with no comparison group
Observational study An association between intake and an outcome in a population Cause and effect; people who take supplements differ in many other ways “Users had 30% lower risk” presented as a product effect
Randomized controlled trial (RCT) Whether the supplement beats placebo for a defined outcome in a defined group Whether results apply beyond the people, dose, and duration studied One small trial cited as if it settled the question
Systematic review and meta-analysis What all eligible trials together say, with quality assessed More than the underlying trials allow; weak trials pooled are still weak A “review” written by the manufacturer that cites only favorable studies

The last two rows deserve the most attention. A randomized controlled trial assigns participants to the supplement or a comparison by chance, which the NIH explains is done to avoid bias; a placebo is “an inactive product that resembles the test product”; and blinding keeps participants and researchers from influencing results, with double-blind trials being the stricter form (NIH). A systematic review, in the Cochrane Library’s definition, “attempts to identify, appraise and synthesize all the empirical evidence” meeting pre-set criteria, and a meta-analysis combines the results into one overall estimate (Cochrane Library). When an independent review of that kind exists for an ingredient, it is usually the best place to start.

Skill 4: Ask five questions about any single study

When a product cites a specific study, these questions separate meaningful findings from noise.

  1. Was it in humans? If not, the claim is about biology, not about you.
  2. How many people, for how long? The NIH’s description of drug trials gives a sense of scale: Phase I safety studies enroll 20 to 80 people, Phase II trials 100 to 300, and Phase III effectiveness trials 1,000 to 3,000. Supplement trials are often far smaller. Small trials are not worthless, but their results are far more likely to be chance or to shrink when repeated.
  3. Was there a comparison group, random assignment, and blinding? The FTC lists control, randomization, double-blinding “when feasible,” validated measures, and statistically significant results as the elements of a substantiating study. A trial where everyone got the product and reported feeling better cannot show an effect.
  4. What exactly was measured, and does it match the claim? A change in a blood marker is not a change in disease risk. The FTC also warns that post hoc subgroup analyses carry an increased risk of false positives: if the headline result appears only in “women over 50 with low baseline levels,” it was probably found after the fact.
  5. Who paid, and has it been repeated? Funding does not invalidate a study, but a single manufacturer-funded trial is where skepticism should be highest. The FTC requires that a claim be “adequate in the context of the surrounding body of evidence,” a formal way of saying one study is rarely enough.

The ginkgo story illustrates why size and design matter. Small studies and lab work long suggested ginkgo might protect the aging brain. Then the Ginkgo Evaluation of Memory study randomized more than 3,000 adults aged 75 or older to ginkgo or placebo for a median of six years and found no difference in rates of dementia or Alzheimer’s disease (NCCIH). Larger, better-designed evidence overturned a plausible hypothesis, which is exactly what it is supposed to do.

Skill 5: Check the research yourself

Four free resources cover most of what a careful reader needs.

PubMed

PubMed is a U.S. government database of citations for more than 32 million scientific papers. NCCIH’s guide suggests searching an ingredient name, narrowing with a condition (“echinacea” plus “colds” cut results from 1,325 to 175 in its example), and using the filters to limit results to randomized controlled trials, free full text, or the past five years (NCCIH). If a product claims to be clinically studied and you cannot find a matching human trial, that is informative.

ClinicalTrials.gov

This registry and results database, maintained by the National Library of Medicine, lists publicly and privately supported clinical studies across all 50 states and 193 countries. Two points matter: the information is supplied by the sponsor or investigator, not vetted by NLM, and registration shows that a trial was planned, not that it worked (NLM). A company “running clinical trials” may be telling the truth and still have no results.

Cochrane and the NIH fact sheets

Cochrane Reviews are systematic reviews published by an independent, not-for-profit network and updated as new evidence appears (Cochrane Library). For individual nutrients and herbs, the NIH Office of Dietary Supplements fact sheets and NCCIH’s “Herbs at a Glance” pages summarize evidence, doses, and safety in plain language; the articles on this site, including our guides on ashwagandha and creatine, lean on these same sources, as our editorial policy explains.

Skill 6: Read the label, not just the claim

Even a well-supported ingredient can be sold in a product that cannot deliver it. Research applies only at the dose and form studied, so compare the Supplement Facts panel against the trial: a proprietary blend that hides amounts, or a dose a fraction of what the research used, breaks the chain between the evidence and the bottle. Our guide to reading a Supplement Facts label covers that comparison, and because the FDA does not test products before sale, third-party certification seals are the main way to check that what is on the label is in the capsule. Evidence of benefit is also only half the question: even well-studied ingredients can interfere with medications, as our guide to supplement–medication interactions shows.

The checklist

Run these in order. A “no” on any of the first four is usually enough to stop.

  1. Is the claim a structure/function statement (asterisk and disclaimer), a qualified health claim (with limiting language), or an implied disease claim the law does not allow on a supplement?
  2. Does the company cite a specific study, and can you find it in PubMed?
  3. Was the study in humans, with a comparison group and random assignment?
  4. Does the outcome measured match the benefit advertised?
  5. How many participants, for how long, and who funded it?
  6. Has an independent systematic review or an NIH fact sheet weighed the whole body of evidence, and what did it conclude?
  7. Does the product deliver the studied dose and form, in a disclosed amount, with third-party testing?
  8. Does the marketing rely on testimonials, “miracle” language, suspiciously exact percentages, or “natural” as a safety argument?

Bottom line

Evaluating supplement claims is a learnable skill, not a matter of trusting the right brand. The law lets companies make structure/function claims without review, the FTC steps in only after deceptive ads have run, and the research behind any given ingredient ranges from a few cell-culture experiments to dozens of human trials. Ask what kind of study, in whom, measuring what, funded by whom, and reviewed by whom, and most claims sort themselves into “supported,” “preliminary,” or “unsupported” quickly. For more on separating evidence from marketing, browse our Supplement Safety & Science section.

Frequently asked questions

Does “FDA registered facility” mean the FDA approved the product?

No. Facility registration is a routine requirement that has nothing to do with whether a product works. The FDA states it does not have authority to approve supplements for safety or effectiveness before sale, and it does not approve labels. Any wording that implies FDA endorsement of a supplement’s benefits is misleading.

What does the asterisk after a claim on a supplement label mean?

It points to the required disclaimer for a structure/function claim: “This statement has not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.” The company must notify the FDA within 30 days of using the claim and must have substantiation on file, but no regulator reviews the claim before it appears.

Is a study on mice or cells ever useful?

Yes, for generating hypotheses and understanding mechanisms. It is simply not evidence that a supplement helps people. The FTC’s guidance states that animal and in vitro studies are not sufficient to substantiate health claims without confirmation in human trials, which is the standard to hold marketing to.

How big does a supplement trial need to be?

There is no single number, but context helps: the NIH describes Phase III drug trials as enrolling 1,000 to 3,000 people, while supplement trials are often much smaller. Small trials can be well designed, but their results are more likely to be chance or to shrink on replication. Look for whether the finding has been repeated and whether a systematic review has pooled the trials.

What is the difference between a systematic review and a regular review article?

A systematic review follows a published protocol to find and appraise all studies meeting set criteria, using explicit methods to limit bias. A narrative review is an author’s summary of studies they chose to discuss, which can be selective. Cochrane Reviews are the best-known systematic reviews in health care.

Can I trust a study if the manufacturer paid for it?

Industry funding does not make a study wrong, but it is a reason to look harder at design, outcome selection, and whether independent researchers have found the same thing. The FTC expects claims to hold up in the context of the surrounding body of evidence, so a single sponsor-funded trial is weak support on its own.

Where should I start if I want to check one ingredient quickly?

The NIH Office of Dietary Supplements fact sheets for vitamins and minerals, and NCCIH’s “Herbs at a Glance” pages for botanicals, summarize the evidence and safety in a few minutes of reading. From there, PubMed with the “randomized controlled trial” filter and the Cochrane Library will show you whether higher-quality evidence exists.

Sources

  1. FDA 101: Dietary Supplements — U.S. Food and Drug Administration, 2022.
  2. 21 CFR 101.93 — Certain types of statements for dietary supplements — Electronic Code of Federal Regulations.
  3. Qualified Health Claims — U.S. Food and Drug Administration, 2024.
  4. Health Products Compliance Guidance — Federal Trade Commission, 2022.
  5. Supplement Marketers Will Relinquish $1.4 Million to Settle FTC Deceptive Advertising Charges — Federal Trade Commission, 2015.
  6. FTC, Maine Attorney General Shut Down Web of Deceptive Supplement Sellers — Federal Trade Commission, 2017.
  7. Natural Doesn’t Necessarily Mean Safer, or Better — National Center for Complementary and Integrative Health.
  8. NIH Clinical Research Trials and You: The Basics — National Institutes of Health, 2025.
  9. About Cochrane Reviews — Cochrane Library.
  10. Ginkgo: Usefulness and Safety — National Center for Complementary and Integrative Health, 2025.
  11. How To Find Information About Complementary and Integrative Health Practices on PubMed — National Center for Complementary and Integrative Health, 2021.
  12. What kind of web site is ClinicalTrials.gov? — National Library of Medicine Customer Support Center.