Skip to content
All posts
Amazon

Health Claims on Amazon Listings: What You Can Say, What Needs Proof, and What Gets You Suspended

Health words are the fastest route from a live listing to an account-level review. The three-bucket rule — allowed, substantiated, regulated — with compliant rewrites for germ-kill claims, 'FDA approved' and 'clinically proven'.

By CopyForge AI 7 min read

Why health words trigger account-level enforcement

Most listing problems are content problems: fix the text, the listing comes back. Health claims are different. The moment your copy says a product cures, kills, treats or protects a body, your general merchandise listing walks into a regulated category — and regulated categories are enforced at the account level, not the listing level.

Amazon's restricted-products policy is plain about it: listings may not claim that a product cures, treats or prevents a disease or condition unless the product is approved for that use. The US regulators behind those words are the FDA (food, drugs, cosmetics, devices) and the EPA (anything that kills or repels a pest, including bacteria on surfaces). You do not need to read either statute end to end. You need to know which of the three buckets your sentence falls into.

Bucket 1 — allowed: cosmetic and general-function language

Words that describe appearance, comfort or general function are yours to use freely, because nobody regulates how soft a towel is:

  • "helps skin feel hydrated"
  • "supports a comfortable night's sleep"
  • "designed for sensitive noses"
  • "keeps food fresh for longer"

Note the pattern: these describe experience and design, not outcomes inside a body. "Hydrated-feeling skin" is a sensation. "Repairs damaged skin barrier" is a physiological claim — and it belongs in bucket 2.

Bucket 2 — needs substantiation: performance and wellness claims

Claims that imply a measurable benefit are allowed only when you hold evidence for the exact claim, on the exact product, before you publish:

  • "clinically proven" — needs a real clinical study of your product
  • "lab tested" — needs the lab report, and the report must test what the sentence claims
  • "reduces bacteria on hands" — needs test data, and often a registration (see bucket 3)
  • "supports immune health" — supplement-structure/function territory: requires substantiation and, in the US, the FDA disclaimer

Two failure modes show up in enforcement again and again. The first is borrowed evidence: the ingredient supplier had a study, so the brand borrowed it. Ingredient studies cover the ingredient in isolation, not your formulation. The second is expired evidence: the study exists, but for an older formulation that is no longer what ships.

Instead of: "Clinically proven to relieve joint pain" Write: "Formulated with 1,500 mg glucosamine per serving"

The rewrite states a formulation fact. A fact about what is in the bottle cannot be an unsupported claim about what happens in a knee.

Bucket 3 — regulated: disease, treatment and pesticide claims

Some words are not a substantiation problem. They are a category problem:

  • "treats eczema", "relieves arthritis", "prevents migraines" — drug claims; your listing is not an approved drug
  • "fights acne at the source" — drug claim about a disease condition
  • "kills 99.9% of germs" — a pesticide claim (section below)
  • "antiviral", "disinfects", "sanitises" — pesticide claims
  • "FDA approved" — almost never true for consumer goods (section below)

If a sentence promises a change in a disease or a body condition, no amount of test data makes it legal for your pillow, bracelet or water bottle. The sentence itself is the violation.

"Kills 99.9% of germs" is an EPA claim, not a marketing one

Under US federal law (FIFRA), any claim that a product or substance kills, controls or repels a pest makes that product a pesticide — and "pest" includes bacteria, viruses and fungi on surfaces. Pesticides must be registered with the EPA before they are sold with those claims, and every efficacy number on the label must come from the registration's approved test data.

This one catches sellers of phone cases, chopping boards, toothbrush holders, yoga mats and "antimicrobial" fabric by surprise. The product is not a spray — but the claim makes it a regulated pesticide device or treated article, with specific allowed wording. "Antimicrobial" alone is a claim that invites the review.

Amazon enforces this hard because it is federally illegal to sell the claim, not just against house rules. That is why entire listings vanish over one bullet point.

Instead of: "Antimicrobial coating kills 99.9% of bacteria" Write: "Smooth, non-porous surface wipes clean in seconds"

If your product genuinely is a registered article, use the EPA-accepted phrasing for your category and keep the registration number in your records. Everyone else: describe cleanability, not killing.

"FDA approved" almost never means what sellers think

The FDA approves new drugs and certain high-risk (Class III) medical devices. It does not approve cosmetics, dietary supplements, food-contact materials, or most consumer goods — and it does not approve listings.

Registered, cleared and approved are three different words: a facility can be FDA registered (an administrative listing that means nothing about product quality), a medium-risk device can be FDA cleared (510(k) substantial equivalence), and almost nothing you sell is FDA approved. Using the wrong one is a false claim even when your paperwork is real.

Instead of: "Made in an FDA approved facility" Write: "Produced in an FDA-registered facility" — only if the registration exists and you can produce it.

Better still: leave regulators out of the copy. No shopper has ever chosen a water bottle because the factory filled in a form.

"Clinically proven" needs the study behind the exact sentence

The enforcement question is never "do you have a study?" It is "does your study test this exact claim, on this exact product, in this exact population, with this exact outcome?" A trial of a different concentration, a different age group, or a different endpoint does not substantiate the sentence you wrote.

If you do hold that study, quote its specifics instead of the adjective: "In a 28-day consumer-perception study of 112 users, 86% said skin looked smoother." Specifics convert better than "clinically proven" ever did, and they survive review.

What CopyForge flags automatically

CopyForge's compliance scan treats these as separate rule families, so a bucket-1 sentence never gets punished for a bucket-3 problem:

  • Disease and treatment claims — "cures", "treats", "prevents", "heals" — flagged as prohibited for unregistered products, per Amazon's restricted-products policy.
  • Pesticide claims — "kills germs", "antibacterial", "sanitises", "disinfects" — flagged with the EPA note so you can decide whether your registration covers the wording.
  • Unsubstantiated proof claims — "clinically proven", "lab tested", "FDA approved" — flagged until you attach the evidence the sentence implies.

Every flag quotes the exact matched phrase so you fix the sentence, not guess at the rule.

The 60-second pre-publish check

Read your listing once and ask, for every sentence:

  1. Does this sentence promise a change inside a body or against a microbe? If yes, is the product registered for exactly that use?
  2. Does it cite a study, a lab, or a regulator? If yes, can I produce that document today?
  3. Does "FDA" appear anywhere? If yes, is the verb registered, not approved — and is the registration mine?
  4. If Amazon asked for evidence this week, would I have it — for this exact wording?

Four confident yeses and the listing is a listing. Anything less is a draft.

A closing honesty note

This is wording-level guidance, not legal advice. Regulated categories — supplements, devices, pesticides, children's products — carry requirements far beyond copy: registration, labelling, testing and record-keeping. What wording-level discipline does do is remove the most common, most avoidable cause of a pulled listing: the sentence you did not need to write.

Keep reading

New rules, explained before they cost you a listing

Occasional email when a marketplace policy changes or we add a rule. No drip sequence or automatic upsell. Every message includes an unsubscribe link.