Testing labs

What independent tests of peptides sold online have found

Researchers, medicines regulators, anti-doping laboratories and journalists have tested peptides bought online or seized at borders. What they found, with every figure linked to its source, and what the evidence can and can't tell you.

What the evidence supports. Products sold online as research peptides are often not what the label says: testers keep finding the wrong molecule, no active ingredient, the wrong amount and, less often, endotoxin, metals or bacteria. The amount in the vial is wrong more often than the identity, in both directions. And a COA shown by a seller is not proof on its own: labs have disowned COAs in their name, and a vendor's owner admitted forging them.

What it can't support. A market-wide failure rate (no study sampled the market at random, and published rates from about 20% to over 70% depend mostly on the pass standard and on which products were sent in); a verdict on a vendor from one test; or anything about safety or effectiveness in people.

Bought online and tested

Semaglutide from illegal online pharmacies (researchers in Hungary and the US, 2024). Of 6 orders, 3 arrived. All 3 vials held more semaglutide than labelled, 29% to 39% more, yet semaglutide made up only 7.7% to 14.4% of the powder against a claimed 99% purity. One vial had 8.95 EU/mg of endotoxin. [1] [2]
Melanotan II from three online shops (Denmark, 2015). Every shop claimed 10 mg per vial; the vials held 4.32 to 8.84 mg, and two shops’ vials had 4.1% to 5.9% unknown impurities. [3]
ACE-031 sold as a research chemical (14 products from UK, European, Chinese and US sellers, 2025). None contained ACE-031: twelve held a different protein, one follistatin 344 and one ipamorelin. [4]
Follistatin 344 from the black market (2019). 9 of 17 products contained follistatin; some of the others held MGF or GHRP-2 instead. [5]
Falsified peptide drugs from three illegal internet pharmacies (Belgium’s medicines control laboratory, 2018). Purity of 5% to 75% for the cysteine-containing peptides, large variation in dose, lead in one sample and arsenic up to ten times the limit for injectable drugs in several. [6]
A 2026 preprint on 6,441 samples of “research grade” peptides sold to consumers reports that 41.6% to 71.1% failed basic quality criteria, depending on the standard applied, and that 15% had measurable endotoxin. It is a preprint, not yet peer reviewed, and we could read only its abstract. [7]

Seized products, tested by regulators and anti-doping labs

Fewer than 20% of 1,190 doping products seized at the Swiss border contained the claimed substance in the claimed amount (2017; all kinds of products, 146 of them labelled as peptide hormones or growth factors). [8]
In Germany, 337 black-market products seized from 2010 to 2013 held 67 different active ingredients; 12.8% were peptide hormones or growth factors. [9]
Europe’s official medicines control laboratories analysed 1,104 suspect products from 2017 to 2019: 11% contained research peptides, the second most reported category. [10]
Wrong or modified molecules keep turning up: glycine-extended versions of GHRP-2, GHRP-6, ipamorelin and modified GRF 1-29 in Danish, Polish and German seizures; a “TB-500” product containing a 7-amino-acid fragment of thymosin beta-4; His-tagged IGF-1 LR3 made for laboratory research; growth hormone vials with no active content; and an unknown fusion protein. [11] [12] [13] [14] [15] [16] [17]
Counterfeit GLP-1 drugs: a German official laboratory found insulin glulisine, not semaglutide, in falsified Ozempic pens, and case reports describe low blood sugar and insulin poisoning from products sold as semaglutide. [18] [19] [20]

Tests paid for by journalists

WIRED (September 2026) had 4 vials bought at a Brooklyn bodega tested: none contained the labelled active ingredient. [21]
The New Yorker (April 2026) had three products from one vendor, Swiss Chems, tested: the BPC-157 contained lead, the TB-500 endotoxin, and the CJC-1295 held under half its labelled amount. A MOTS-c vial it also had tested was high purity but below its labelled amount. [22]
New York Magazine (January 2026) had a retatrutide vial ordered from China tested: it appeared to be the right compound, correctly dosed, with mannitol as filler. [23]

Each is one purchase per product: a snapshot, not a verdict on the seller.

What testing platforms report

Finnrick, which publishes tests of samples buyers send in, told CNN in April 2026 that about one in five of almost 9,900 tests failed a looser standard (the right compound, purity of at least 98%, and quantity within 20% of the label), and that quantity was the most common problem. Other outlets reported “about a third” and “nearly 30%”: failure rates depend on the standard applied and on which samples buyers chose to send. [24] [25] [23]
Endotoxin: Finnrick found a quantifiable amount in about 1 in 12 of the 140 samples tested for it (February 2026). Endotoxin testing was an optional paid add-on, so this is 8% of a self-selected group, not of all samples, as some reports put it. [26] [27] [28]
Labs get things wrong too: Finnrick’s corrections log shows lab results revised in both directions (a buffer peak read as an impurity, 92 mg entered for 9.2 mg), and a calibration error that put the same wrong potency on five reports. [29] [30]

Finnrick's figures are its own data, as it or the outlets named published them; we haven't recalculated them.

Forged and disowned COAs: the documented cases

2026-10-01 FDA says it knows of fraudulent compounded semaglutide and tirzepatide whose labels name pharmacies that don’t exist, or licensed pharmacies that didn’t make the product. A product label, not a COA, but the same false use of a real business’s name. [31]
2026-07-30 The owner of Paradigm Peptides was sentenced to 5 years 10 months in prison after pleading guilty; the Justice Department says many of its products sold as SARMs contained testosterone. CBS News reports he admitted forging laboratory certificates for products advertised as 99% pure and third-party tested. [32]
2026-07-02 CBS News reported that COAs on INDR Labs’ website named Vanguard Laboratory, and that the lab said it hadn’t issued them and had no client called INDR Labs; one “retatrutide” COA appears to show tirzepatide. WIRED’s own tests of four INDR Labs vials later found none of the labelled ingredients. [33]
2026-03-26 Finnrick retracted all its published retatrutide and tirzepatide tests for one Chinese vendor, citing “clear signals of large-scale fraud” in its name and a site carrying “an obviously fraudulent copy of the Finnrick web site”. [29]
2026-03-01 Finnrick wrote that COAs can be edited after issuance, or created without any lab, and that it has “seen multiple examples of doctored COAs being distributed”. [34]
2025-11-14 Finnrick removed retatrutide results it had published under Peptide Sciences’ name: the sample was a counterfeit. Products sold under a known vendor’s name aren’t always that vendor’s. [29]
2025-03-19 TrustPointe reported its first case of a business falsifying its COAs: some results weren’t even possible values for the test, and the QR code and verification link had been removed every time. [35]
2020-05-14 In dietary supplements, Alkemist Labs reported suppliers altering its valid reports: changing client names, replacing lot numbers and, in some cases, results. The closest documented precedent outside peptides. [36]

We also looked for vendors publishing another vendor's COAs in 105 saved vendor COA pages and found shared files only between two apparent sister sites. What you can do: look a COA up with the lab that issued it, check what it says, and check the lab is real.

Independent testing: questions people ask

Are research peptides sold online what the label says?

Often not. Researchers, medicines regulators, anti-doping labs and journalists have all found products with the wrong molecule, no active ingredient, the wrong amount and, less often, endotoxin, metals or bacteria. No study has sampled the market at random, so there is no reliable overall failure rate; the figures on this page say which products each one tested.

What goes wrong most often when peptides are tested?

In the largest datasets, the amount in the vial: Finnrick told CNN that quantity was the most common problem. Errors run both ways; the semaglutide vials in one study were overfilled by up to 39% yet held mostly other material. Purity alone doesn’t show this, because HPLC purity is a share of the detected peaks, not of the powder.

Can a COA be forged?

Yes. Labs have disowned COAs carrying their name, a vendor’s owner admitted forging certificates in a federal case, and testing platforms report doctored COAs. Check a COA with the lab that issued it: how to verify a COA at each lab.

What does this evidence not show?

It doesn’t give a market-wide failure rate, a verdict on any vendor from a single test, or anything about safety or effectiveness for people. Tests describe the vials tested; batches differ, and labs make errors too.

Sources

Compiled 2026-10-16 from sources we opened. Papers we couldn't read beyond a title are left out. Nothing here is advice on using any product. See also: what testing costs · testing labs compared.