State of research peptide quality, 2026

What published studies, regulator tests and large testing datasets show about peptides sold online outside the regulated drug supply, and what they cannot show.

Last reviewed 2026-10-25 by the PeptideWiki editorial team. Editorial policy.

Key findings

• Identity is the first failure point. In an Australian laboratory survey of more than 200 products, only about a quarter contained the peptide on the label. Two regulators separately found semaglutide-labelled products with no semaglutide, and pens that contained insulin.
• Illicit-channel purity can be extremely low. Semaglutide vials bought online measured roughly 8–14% pure against a 99% claim; falsified peptides in Belgium ranged down to about 5%.
• Amount varies in both directions. Melanotan II vials held 43–88% of the label; online semaglutide vials held 29–39% more than labelled.
• Contaminants go beyond peptide impurities. Studies found endotoxin in every tested semaglutide vial, inorganic arsenic and lead, and a pathogenic Bacillus strain.
• Some products are new molecules. Anti-doping labs found growth hormone and GHRP variants with extra residues: compounds with no safety or pharmacology data at all.
• Community testing data are better but not reassuring. In about 6,400 lab reports, median purity was about 99.8%, yet roughly 2.4% of samples lacked the named peptide, and around 42% would fail compounding-style specifications. Endotoxin was measured in under 4% of samples.

Identity failures by compound

From the largest published analysis of third-party lab reports on consumer-marketed research peptides (a non-peer-reviewed preprint; samples are not a random draw from the market):

How many samples would meet pharmaceutical specifications

The same dataset judged against two sets of limits: compounding-style (90–110% of label, at least 98% pure) and manufactured-drug style (95–105%, at least 99.5% pure).

A second, larger pool of about 64,800 community lab records used looser thresholds and found outright identity failures in under 1% of tests, with more than 10% underfill in about 7% of quantity tests. The difference between the two analyses comes mostly from which limits are applied, which is itself a reminder that "passes testing" depends on the standard used.

What the evidence cannot tell you

• Representativeness. Community-testing samples are submitted by vendors or buyers and are likely skewed toward better products; regulator and seizure samples are skewed toward the worst. Neither is a random sample of the market.
• Sterility and endotoxin. This is the largest gap. Fewer than 4% of community-tested samples had an endotoxin result, and sterility testing is almost absent, even though these matter more than a fraction of a percent of chemical purity for any cell-based work.
• Lot-to-lot consistency. A good result for one lot says little about the next.
• Age of the evidence. Several analytical studies are from 2014–2018; the GLP-1 findings are recent.

What this means for laboratories

1. Treat identity as unproven until mass spectrometry confirms it for your lot. The COA checker compares a reported mass with the expected ions.
2. Ask for the chromatogram and spectrum, not just a purity number, and check that the COA lot matches the vial.
3. For cell-based or in-vivo work, require an endotoxin result, or test it yourself; see endotoxin testing.
4. Use net peptide content, not powder weight, to set concentrations; see purity vs net peptide content.
5. Consider independent re-testing for any lot that matters to your results.

Evidence reviewed

18 sources: peer-reviewed studies, regulator testing and alerts, and testing-program data. Summaries are in our words; follow the links for full details. Where a source reported on a specific seller, we do not name it.

Method

We searched for analytical studies, regulator test results and testing programs reporting on peptides sold outside the regulated drug supply, opened each source, and recorded what was tested and the quantitative findings. Vendor marketing and vendor-run reviews were excluded. Two sources are not peer-reviewed (a preprint and an independent data aggregate) and are labelled as such. We will update this report as new evidence is published.

Disclosure: PeptideWiki is owned by a company that also operates research-peptide stores; see the disclosure. No store or vendor is ranked or named in this report.