How can you tell if a supplement is pure, accurately labelled, and safe?

Nobody reviews quality before sale. The FDA does not review supplements for safety before they reach shelves¹, and a public-health review concluded that enforcement challenges remain large¹⁰.

How quality fails. Audits keep finding four problems:

• The wrong amount. Twenty-eight red yeast rice brands held from none to 5.48 mg of the statin-like compound monacolin K per 1,200 mg, a 60-fold range among those that contained it⁹. See our red yeast rice page.
• Toxic metals. One in five Ayurvedic medicines bought online carried detectable lead, mercury or arsenic⁸. For ordinary mineral and multivitamin products, lead levels have fallen since 2000, but some outliers remain².
• Hidden drugs. The FDA warned about 776 supplements containing undeclared drugs over ten years, mostly sexual-performance, weight-loss and muscle products³.
• Banned sports substances. About 15% of 634 non-hormonal supplements bought in 13 countries held undeclared anabolic steroids⁵.

What seals can and cannot do. Third-party programs test products or audit factories for identity, strength and contaminants, and programs built for athletes also screen for banned substances. They differ in what they test, so check what a seal covers¹. USP and NSF seals indicate low lead levels². No seal shows that a supplement works.

A practical checklist

1. Look for a seal, then confirm on the certifier's own website that your exact product is listed. A logo on a label is not proof.
2. Treat any product that promises drug-like effects, especially for weight loss, sex or muscle gain, as suspect³.
3. Search the FDA's Health Fraud Product Database. It lists only a small fraction of risky products, so absence from it is not clearance⁴.
4. Ask for a batch-specific certificate of analysis from an independent lab. A generic one proves little.
5. If you take medicines, ask a pharmacist, and read which herbs interact with medications.

New to supplements? Our multivitamin guide covers whether you need one at all.

Toxic metals

Saper and colleagues randomly selected 230 of 673 Ayurvedic products sold on 25 websites, bought them and analyzed the 193 they received by x-ray fluorescence. Of these, 20.7% (95% CI 15.2% to 27.1%) contained detectable lead, mercury or arsenic, with no difference between US-made (21.7%) and Indian-made (19.5%) products. Rasa shastra products, which deliberately combine herbs with metals, were worse (40.6% versus 17.1%). Of the products containing metals, 75% claimed good manufacturing practices, and every one exceeded at least one daily-intake standard⁸. Ordinary supplements fare better. A narrative review found calcium from dolomite or bone meal historically carried more lead than refined forms, but the gap has narrowed since 2000, and multivitamin-minerals assessed since 2007 had low average lead, though with large differences between products².

Hidden drugs

Tucker and colleagues analyzed FDA warnings from 2007 to 2016. They counted 776 adulterated supplements from 146 companies: 45.5% marketed for sexual enhancement, 40.9% for weight loss and 11.9% for muscle building, and 20.2% had more than one hidden drug. Sildenafil was in 47.0% of the sexual products, sibutramine in 84.9% of the weight-loss products and steroids in 89.1% of the muscle products. Of 28 products named in more than one warning, 19 (67.9%) were reported to hold new unapproved ingredients in a later warning. In 2014 to 2016, 38.6% of adulterated samples were found by online sampling and 34.3% in international mail shipments³. This counts regulatory warnings, not the share of all products on shelves.

Dose accuracy

Cohen and colleagues analyzed 28 red yeast rice brands from mainstream US retailers. Monacolin K was undetectable in two. In the other 26 it ranged from 0.09 to 5.48 mg per 1,200 mg, and at the manufacturers' serving suggestions the daily dose ranged more than 120-fold, from 0.09 to 10.94 mg⁹. FDA manufacturing standards did not produce predictable strength.

Sports supplements

Geyer and colleagues bought 634 non-hormonal supplements in 13 countries in 2000 to 2001. Of these, 94 (14.8%) held anabolic steroids that the label did not declare, including 18.8% of those bought in the USA. Rates were 21.1% for companies that also sold prohormones and 9.6% for those that did not. Taking supplements that contained nandrolone prohormones, more than 1 microgram in total, produced positive doping tests for norandrosterone for several hours⁵. The data are two decades old, but they are why athlete-focused certification programs exist.

Herbal identity and the DNA-barcoding story

A widely quoted 2013 DNA-barcoding survey of North American herbal products reported extensive substitution, but BMC Medicine retracted it in 2024, so we do not use its figures¹¹. A smaller sequencing study of 15 products from three producers found DNA from the labeled plant in 7 of 8 dry-herb supplements. In extracts, the DNA was mostly too degraded to read: the labeled plant was detected in only 3 of 7 extracts by next-generation sequencing and in none by Sanger sequencing, yet chemical analysis still found ten key ginkgo compounds in samples with degraded DNA¹². DNA tests can show a plant is missing, but "no DNA" does not prove an extract is fake. Chemical testing is also needed.

Harm reports

Surveillance data estimated 23,005 emergency department visits and 2,154 hospitalizations a year for supplement adverse events in the US (3,667 cases, 2004 to 2013). Herbal and complementary products for weight loss (25.5%) and for increased energy (10.0%) were commonly implicated, and the two together caused 71.8% of the palpitation, chest-pain or rapid-heartbeat events⁷. In the Drug-Induced Liver Injury Network, 130 of 839 cases (15.5%) were attributed to herbal and dietary supplements, rising from 7% to 20% of cases over the study period. Non-bodybuilding supplements led to death or transplant more often than medications did (13% versus 3%)⁶. These are case series and surveillance estimates, not rates per user.

Seals and their limits

A consensus statement notes that no regulator conducts or mandates a quality review before sale, that third-party programs have varying testing components, and that consumers find it hard to judge whether a given program lowers risk enough¹. We found no study comparing harm from certified and uncertified products, so the case for seals rests on what they test, not on proven outcomes.

Bottom line (confidence 3 of 5). Quality failures are real, repeated across independent audits, and concentrated in particular product types. Seals are the most practical safeguard, but direct proof that they prevent harm is limited.