Before an herb becomes a capsule, someone must prove it’s the right plant

Key takeaways

• Under rules finalized in 2007, a supplement company has to verify with at least one test or examination, not a supplier’s word alone, that each herb or other dietary ingredient is what it should be before it goes into a product, unless the FDA grants an exemption³⁴.
• Products sold as tejocote root turned out to contain yellow oleander, a poisonous plant, and the FDA was still adding products to its warning in September 2026¹².
• A testing company and a consultant who both serve the supplement industry list the identity-testing rule among the findings FDA inspectors cite most often, though neither publishes counts⁵⁶.
• No single test works for every herb: DNA tests often fail on extracts, and some chemical tests can be fooled by cheap added compounds¹²¹³.
• Ask a brand which identity test it ran, what reference it compared against, and whether the test rules out the herb’s known look-alikes.

In September 2022, an emergency physician called New Jersey’s poison center about a toddler, not yet 2, who had swallowed a product labeled Mexican tejocote root². The child’s heart rate and blood pressure had dropped. Doctors gave the antidote used for digoxin poisoning, and a second dose when the heart effects came back².

Tejocote is a Mexican hawthorn whose root is sold for weight loss¹². That December, the poison center, based at Rutgers, bought 10 tejocote weight-loss products online and shipped them to a private testing lab².

None of the 10 contained tejocote. Nine contained yellow oleander².

Yellow oleander is a poisonous plant native to Mexico and Central America¹. It holds heart-active compounds that act like the heart drug digoxin, the team explained in the CDC’s Morbidity and Mortality Weekly Report in September 2023². The FDA says the plant can cause nerve, gut and heart effects that may be severe or fatal¹.

The FDA then sampled and tested more tejocote products¹. The warning kept growing. By September 2026, it listed more than two dozen products sold through Amazon, eBay, Etsy and brand websites, some recalled and some whose sellers the FDA could not reach¹.

Federal rules require a basic identity check, run before the powder goes into capsules, that should catch exactly this. Here is how that check works, where it fails, and how to tell which companies take it seriously.

The rule: test every herbal ingredient before it’s used

The FDA’s manufacturing rules for supplements, known as current good manufacturing practices, were finalized on June 25, 2007⁴. Larger companies had to comply by June 2008. The smallest, with fewer than 20 full-time workers, had until June 2010⁴.

The key line sits in 21 CFR 111.75. It says each dietary ingredient’s identity has to be verified by at least one appropriate test or examination, and that happens before the ingredient is used³. “Dietary ingredient” covers the herb, vitamin or other active part of the product.

The FDA explained why in 2007. It called dietary ingredients the central, defining ingredients of a supplement⁴. If the herb is wrong, nothing else about the product can be right.

Three details matter for shoppers:

• A supplier’s paperwork isn’t enough for the herb itself. For other components, such as fillers, a company can lean on the supplier’s certificate of analysis, a lab report, once it has confirmed that supplier’s results are trustworthy³. The dietary ingredient needs an actual test or examination that the company is responsible for³.
• The test must be scientifically valid. The rule lists acceptable kinds: sensory checks, examination by eye, microscopy, chemical analysis or another valid method³.
• Exemptions exist, but they take data. A company can petition the FDA to test less often, but it must show with data, such as 95% statistical confidence that a bad lot is rare, that assurance won’t meaningfully drop⁴.

Our guide to comparing supplement brands covers the rest of these manufacturing rules.

Inspectors keep finding the identity step missing

How often do companies skip it? The most readable summaries come from industry, and they come without numbers.

Eurofins, a company that sells lab testing, summarized FDA inspection findings since 2009. The most frequent citations fell in one part of the rules, Subpart E, and the identity-testing requirement made Eurofins’ top five from that part. The undated page gives no counts or rank order⁵.

Gary Callahan, a consultant to supplement and drug makers since 2007, wrote in the trade journal Nutritional Outlook in 2025 that identity testing is one of the five FDA findings he sees most often. He added that many firms still release an ingredient on a supplier’s certificate alone⁶.

Both sources sell services to supplement makers, so read them as informed views, not official tallies. But FDA warning letters show what the gap looks like.

In 2024, the FDA sent a warning letter to Restorative Botanicals, a Colorado maker of mushroom gummies and capsules⁷. The company had identified chaga, maitake and Agaricus extracts by color, smell and taste. The FDA said that wasn’t good enough, because all three can show up as similar brown powders⁷.

Its test of the finished products had the same flaw. With several botanicals blended, their signals overlapped into one profile that couldn’t tell them apart⁷.

In 2018, the FDA cited Jinher, Inc., a California supplement maker⁸. The company planned to check new ingredients against the last batch it had tested in-house, using an infrared scanner. The FDA said comparing new lots with earlier lots “would not be considered a scientifically valid method”⁸. A real check needs a qualified reference standard⁸.

That second case shows a quiet trap. Matching your last delivery proves consistency. It doesn’t prove the plant was ever right.

How labs tell one plant from another

Herbalists and botanists have long known plants by their roots, leaves, smell and taste. Labs keep those skills and add instruments, because a capsule hides almost everything an expert would look at.

The U.S. Pharmacopeia (USP), a standards body for medicines and supplements, sets out botanical identification in its General Chapter 563⁹. Its training stresses two ideas: examine the plant by eye and under the microscope, and confirm with chemical tests⁹. It calls for orthogonal methods, meaning tests that check different things, so one method’s blind spot is covered by another⁹.

Here is how the main methods compare.

Method	What the lab looks at	Best use	Blind spot
Sensory check	Look, smell, taste, feel	Whole or cut herbs with clear traits3	Look-alike powders; FDA rejected it for three brown mushroom extracts7
Examination by eye	Shape, color and structure of plant parts	Whole or coarsely cut raw herbs9	Little left to see once a plant is finely ground
Microscopy	Cell types, hairs, crystals, starch grains	Raw herbs, including powders9	Less useful after a plant is dissolved into an extract
HPTLC fingerprint	A pattern of bands on a coated plate, compared with a reference	Comparing closely related plants10	Only as good as the reference sample used810
HPLC marker test	Amounts of chosen compounds	Checking strength and key markers9	Cheap pure compounds can make totals look right13
Broad mass spectrometry	A wide chemical profile	Catching unexpected additions; telling skullcap from germander1214	Complex; worked better on leaf than extract in one NIST study12
DNA barcoding	A short stretch of genetic code	Raw plant before processing13	Extracts often lack usable DNA; can’t detect dyes or added chemicals13

HPTLC stands for high-performance thin-layer chromatography. The lab spots a sample and reference samples side by side on a coated plate. A solvent carries each sample’s compounds up the plate, leaving bands.

Under UV or white light, the analyst compares the bands’ position, color and strength with the official description¹⁰. USP’s chapter on the method (a 2015 edition, posted on a third-party site) says its reproducible results let labs compare closely related plant materials¹⁰.

HPLC, or high-performance liquid chromatography, measures how much of particular compounds a sample holds. Labs often use it to check strength. But it answers only the question it’s asked.

A test is only as good as the reference beside it

Every identity test is a comparison. The sample is checked against something known to be right.

USP issues reference standards for this, and its training covers how to use them in botanical identification⁹. The HPTLC chapter builds them into each run: two or more known substances must separate as expected before results count¹⁰.

The National Institute of Standards and Technology (NIST) adds another layer. In 2007, with partial funding from the NIH Office of Dietary Supplements and an FDA drug center, it released standard reference materials for ginkgo leaves, ginkgo extract and ginkgo tablets¹¹. They came with certified values for key plant compounds and for four toxic elements, including lead and mercury¹¹.

NIST has since built similar materials for herbs including saw palmetto, green tea, St. John’s wort, turmeric, ginseng and bilberry extract¹¹. Sets based on black cohosh are in progress¹¹.

NIST says these materials are meant mainly for developing test methods and as quality controls¹¹. They help a lab prove its test works. They don’t replace testing each incoming lot.

Botanists add one more safeguard for raw plants: the voucher, a preserved specimen that lets others double-check a name. Our article on plant look-alikes explains how vouchers work.

Why extracts are the hardest case

A dried leaf still has its cells, its DNA and its full chemical makeup. An extract has been soaked, filtered, concentrated and often dried with carriers. Most of the clues are gone.

DNA breaks down. Staff of the American Botanical Council, an industry-supported nonprofit, reviewed published testing in 2024¹³. They noted that heavily processed herbs often yield no DNA that can be read. So a missing DNA signal doesn’t prove fraud, and a positive one doesn’t prove the product is authentic¹³. DNA tests also can’t spot the wrong plant part, an added dye or a pure chemical¹³. Our plant identification article covers the New York case that made this limit famous.

Chemical totals can be gamed. Ginkgo extracts are standardized to a set level of flavonol glycosides, a family of plant compounds. Adulterators add cheap pure flavonoids such as rutin and quercetin¹³. A test that only totals that family may miss the swap. The same review found that a simple light-absorption test is often the only check for elder berry extracts, so look-alike pigments can slip past¹³.

Even advanced methods struggle. NIST chemists spiked ginkgo samples in the lab and ran them through broad mass spectrometry¹². For ginkgo leaf material, the analysis roughly sorted samples into unspiked, lightly spiked and heavily spiked groups, driven by markers of Japanese sophora, a common adulterant¹². For dried leaf extracts, it couldn’t separate the spiking levels across the full set of samples. It worked only within material from the same source¹². The authors suggested the method suits investigations better than routine screening¹².

That’s why the council’s review concludes that most botanicals may need several complementary methods, each with its own limits¹³. It’s also why the best time to confirm an herb is before it’s extracted, while the plant still looks, smells and tests like itself.

Wrong plant, real harm: germander and yellow oleander

Most identity failures cheat buyers. A few poison them.

Germander, a group of mint-family plants, was sold in France as a weight-loss extract. By 1992, French doctors were publishing reports of liver damage in people who took it¹⁴.

A 1999 tally found at least 35 suspected cases and 2 deaths, and a 2013 review put the total above 50¹⁴. Several European countries then barred germander as an herbal medicine¹⁴.

NIH’s LiverTox database also notes that skullcap extracts have sometimes been contaminated with germander, and it cites a 2009 study in which mass spectrometry could distinguish the two¹⁴. So the right test, run on every lot, can protect skullcap users from a known liver hazard.

In the reported cases, liver injury appeared anywhere from 2 to 18 weeks after people started germander. It resembled viral hepatitis: people felt tired and nauseated, and their skin turned yellow¹⁴. Yellowing is a sign of liver injury¹⁴, so if you notice it in your skin or eyes while taking skullcap or another herb, see a clinician promptly and bring the bottle.

Yellow oleander is the sharper danger. If you bought a product sold as tejocote root or “Brazil seed,” the FDA says to stop using it, throw it away and contact a health care provider, even if you haven’t taken it recently¹. Call 911 if someone collapses, has a seizure, has trouble breathing or can’t be woken¹⁵. For other poisoning questions, call Poison Control at 1-800-222-1222¹⁵.

Seven questions to ask a supplement company about identity testing

Herbalists have always known that a remedy is only as good as the plant behind it. Asking a company how it confirmed the plant carries that respect into the supply chain, and you’re entitled to a clear answer.

Send these by email so you have the replies in writing.

1. “Which test did you use to confirm the identity of this herb?” A good answer names a method, such as microscopy, HPTLC or DNA testing of the raw plant. A weak answer is “our supplier certifies it,” because the rule requires a real test or examination of the herb, not just the supplier’s paperwork³.
2. “What did you compare it against?” Look for a recognized reference standard or an authenticated sample. Comparing with the last shipment isn’t valid, as the FDA told one company⁸.
3. “Did you test the raw plant or the extract?” Testing before extraction is easier to trust, since processing destroys many clues¹²¹³.
4. “Does your test rule out this herb’s known look-alikes?” For skullcap, ask about germander¹⁴. For tejocote, ask about yellow oleander¹.
5. “Do you use more than one method?” USP stresses tests that check different things⁹.
6. “Can I see the identity result for my lot number?” Many companies won’t share it, but the ones that do are showing their work.
7. “Does an independent certifier check identity for this product?” If so, confirm the product on the certifier’s own website. Our articles on third-party testing and certifications explain what different seals cover.

How to read the answers. Specific methods and references mean a company has a system. Vague words like “rigorous,” “pure” or “lab-tested,” with no method named, mean you should keep asking or buy elsewhere.

What the evidence shows

Claim	Evidence rating	What the research shows
Wrong-plant substitutions reach U.S. shelves and can be dangerous	Strong evidence	Tejocote products held yellow oleander; germander has caused liver injury and contaminated skullcap1214
Sight, smell and taste can confirm a powdered extract	Evidence says no, or unsafe	The FDA rejected that approach for look-alike mushroom powders7
DNA barcoding can confirm any herbal extract	Evidence says no, or unsafe	Processed material often has no readable DNA; testing works best on raw plant13
One test method is enough for every herb	Evidence says no, or unsafe	USP and the council’s review call for complementary methods913
Broad chemical fingerprinting can catch spiked ginkgo extracts	Promising but early	It worked on leaf material but not reliably across extracts in a NIST study12
A brand that says “lab-tested” runs proper identity tests on every lot	Not enough evidence	Industry summaries list identity testing among the most-cited inspection findings, and we found no data linking such claims to actual practice56

Questions readers ask

The bottom line

Before an herb becomes a capsule, federal rules require the maker to prove it’s the right plant. That check sounds basic, yet industry summaries of inspection records place it among the most frequent failures. The tejocote products sold with yellow oleander inside show what can happen when it’s skipped.

No single method can identify every herb in every form. Good companies test the raw plant, compare it with a trusted reference, use more than one method and screen for known look-alikes. Ask them how they do it. A company that answers in detail is giving you the same respect you give the plants.

Related reading

• How experts tell deadly plants from their edible look-alikes
• Third-party tested says what’s in the bottle, not that it works
• How to tell a real certification from a logo
• A supplement brand can outsource the making, not the responsibility
• Some herbal pills hide prescription drugs, and others swap out the herb

Sources

1. U.S. Food and Drug Administration. “FDA Issues Warning About Certain Products Containing Toxic Yellow Oleander.” FDA safety alert, content current as of September 22, 2026. fda.gov
2. Berland N, Santos C, Calello DP, Kababick J. “Notes From the Field: Online Weight Loss Supplements Labeled as Tejocote (Crataegus mexicana) Root, Substituted With Yellow Oleander (Cascabela thevetia) — United States, 2022.” MMWR Morbidity and Mortality Weekly Report, CDC, September 15, 2023;72(37):1016–1017. cdc.gov
3. Electronic Code of Federal Regulations. “21 CFR 111.75: What Must You Do to Determine Whether Specifications Are Met?” National Archives, up to date as of October 15, 2026. ecfr.gov
4. U.S. Food and Drug Administration. “Current Good Manufacturing Practice in Manufacturing, Packaging, Labeling, or Holding Operations for Dietary Supplements; Petition to Request an Exemption From 100 Percent Identity Testing of Dietary Ingredients” (interim final rule). Federal Register, June 25, 2007;72(121):34959–34969. govinfo.gov
5. Eurofins US. “Dietary Supplement Enforcement Trends: What Is FDA Looking For?” Eurofins (a commercial testing laboratory; raw data credited to Marian Boardley Consulting LLC and the FDA data dashboard), undated, accessed October 2026.
6. Callahan G (GMP Consulting Services, Inc.). “Five Most Common GMP Observations by FDA.” Nutritional Outlook, April 24, 2025 (page metadata shows October 19, 2026). nutritionaloutlook.com
7. U.S. Food and Drug Administration. Warning Letter to Restorative Botanicals, LLC (CMS 674740). FDA, August 13, 2024. fda.gov
8. U.S. Food and Drug Administration. Warning Letter to Jinher, Inc. (557722). FDA, October 4, 2018. fda.gov
9. U.S. Pharmacopeia. “Botanical ID Methodology Tests (On-Demand)” (course description covering General Chapters 563 and 203). USP, accessed October 2026. usp.org
10. U.S. Pharmacopeia. General Chapter 203, “High-Performance Thin-Layer Chromatography Procedure for Identification of Articles of Botanical Origin.” USP38–NF33, Supplement 1, 2015 (text as reproduced on the third-party site DrugFuture, not on usp.org; check against the current USP–NF). drugfuture.com
11. National Institute of Standards and Technology. “Measurements and Standards for Botanical Dietary Supplements.” NIST, last updated March 26, 2025; and “Ginkgo SRMs: Tools for Product Analysis/Quality,” August 16, 2007. nist.gov
12. Cruz MB, Place BJ, Wood LJ, Urbas A, Wasik A, Rocha WFC. “A Nontargeted Approach to Determine the Authenticity of Ginkgo biloba L. Plant Materials and Dried Leaf Extracts by Liquid Chromatography-High Resolution Mass Spectrometry (LC-HRMS) and Chemometrics.” NIST author manuscript. tsapps.nist.gov
13. Orhan N, Gafner S, Blumenthal M. “Estimating the Extent of Adulteration of the Popular Herbs Black Cohosh, Echinacea, Elder Berry, Ginkgo, and Turmeric: Its Challenges and Limitations.” Natural Product Reports, 2024 (funded by the ABC-AHP-NCNPR Botanical Adulterants Prevention Program, which is supported by industry members). umb.herbalgram.org
14. National Institutes of Health. “Germander.” LiverTox: Clinical and Research Information on Drug-Induced Liver Injury, last updated March 12, 2018. ncbi.nlm.nih.gov
15. America’s Poison Centers. “Poison Help” (1-800-222-1222). Accessed October 2026. poisonhelp.org