Claims on this page checked 2026-08-20

Research peptides: what is actually in the vial

Research peptides can be close to the label on tested measures. A large submitted-sample study found median purity of 99.80% (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). The same paper found 2.4% of screened reports lacked the named peptide (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). The preprint reports no outside funding.

Those results do not tell you what is in an untested vial. A lab can test the contents of a sample. A name on a label cannot stand in for that test.

The median amount was 101.80% of the label claim (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). The table below gives the sample sizes and spread. These are reports that people chose to send to a test service, not a random market sample.

A separate purchase study received three semaglutide vials (PMID 39509151; checked 20 August 2026). It found semaglutide in them. Their purity results ranged from 7.70% to 14.37% (PMID 39509151; checked 20 August 2026). Endotoxin was found in all three (PMID 39509151; checked 20 August 2026). The Hungarian Scientific Research Fund funded that study.

A test purchase shows what arrived in that purchase. Reports from official labs show what those labs tested. Each type of evidence has a different sample. None is a test of every vial on sale.

What did the largest testing dataset find?

Mendias and Awan used reports sent to Finnrick Analytics. Consumers sent samples. Peptide-making firms sent samples too. Those choices make the data self-selected. The preprint reports no outside funding. Its authors report no conflicts of interest.

Its exact title is Evaluation of Research Grade Peptides Marketed Directly to Consumers Reveals Extensive Variability in Purity and Measured Abundance. The primary identifier is DOI 10.20944/preprints202604.1748.v1, checked 5 September 2026.

Was the named peptide present?

The named peptide was absent in 156 screened reports (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). That is the paper's 2.4% identity-failure figure (same DOI; checked 5 September 2026).

The paper's highest identity-failure rate was for TB-500. That finding rests on four reports (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). The CJC-1295 result rests on 22 (same DOI; checked 5 September 2026). The counts give the scale of each rate. These counts describe the tested reports. They are not counts of all goods on sale.

How close were purity and amount to the label?

The medians were close to the label. The median is the middle result when the values are put in order. It is not a promise about each sample. The IQR spans the middle half of those values. The table gives that spread as well as the middle value.

The paper also tested how many samples met two sets of limits. A sample had to meet both the amount and purity limits to pass.

Under the compounding model, 32.8% met the purity limit but missed the amount window (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). 3.3% failed both tests (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026).

The largest failure group in that model concerned the amount in the vial. It was not the group in which a different named drug was found. The model failure rates do not mean that those shares held the wrong peptide.

What did the dataset report about endotoxin?

The regression used 36 samples with quantifiable endotoxin (DOI 10.20944/preprints202604.1748.v1; checked 2026-09-05). The authors reported no correlation between purity and endotoxin in that group. Their model returned R2 < 0.01 (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026). The P value was 0.75 (DOI 10.20944/preprints202604.1748.v1; checked 5 September 2026).

Purity is not a test for endotoxin. These tests do not ask the same question. The certificate of analysis page sets out what a lab report can prove.

What are the limits of this dataset?

This is a preprint, not a peer-reviewed paper. Its sample is not random. The authors say firms that trust their products may be more likely to send samples. Poor batches may be less likely to enter the data.

The study tested purity. It measured the amount of peptide. It also tested endotoxin in a subset. The panel did not test sterility. It did not test solvents left from the process. No test for particles was run. Stability was not tested. Nor was structural integrity. The authors call their failure rates a lower bound on the true quality deficit. These limits come from the same DOI, checked 5 September 2026.

Both points matter. The samples may lean toward goods the senders thought would do well. The test panel also leaves some drug-quality checks unanswered.

What did the semaglutide test purchase find?

The team ordered from six illegal online pharmacies (PMID 39509151; checked 20 August 2026). Three vials arrived (PMID 39509151; checked 20 August 2026). The other three orders were non-delivery scams (PMID 39509151; checked 20 August 2026).

The Hungarian Scientific Research Fund supported this work through grant 143684 (PMID 39509151; funding checked 21 August 2026). The team used liquid chromatography-mass spectrometry to test the samples. It also tested for viable microorganisms. It ran endotoxin tests too.

The full report is in Journal of Medical Internet Research, DOI 10.2196/65440. A second paper is in JAMA Network Open, PMID 39093567, DOI 10.1001/jamanetworkopen.2024.28280. These are two papers on one purchase study. They are not two independent tests of the market. Both sources were checked on 20 August 2026.

Endotoxin was found in all three vials (PMID 39509151; checked 20 August 2026). One vial reached 8.95 EU/mg, a level the authors called elevated (PMID 39509151; checked 20 August 2026). No viable microorganisms were found in the freeze-dried samples when tested (PMID 39509151; checked 20 August 2026).

Each sample's chromatogram showed one signal (PMID 39509151; checked 20 August 2026). The authors read this as the exclusive presence of semaglutide. The low purity results did not show insulin substitution. The insulin cases below concern different products.

Removing endotoxin is a different process from killing bacteria. Rashed and colleagues call it depyrogenation (PMID 40122208, DOI 10.1016/j.xphs.2025.103769; checked 20 August 2026).

Those are the paper's test conditions, not a method for preparing a vial. The separate sterility and infection page covers that distinction.

What did official laboratories find in seized products?

The GEON study drew on reports from official medicines control labs. Its scope was SARMs, metabolic modulators and growth hormone secretagogues. It was not a survey of peptide vials on the market.

The source is Mendoza Barrios and colleagues, PMID 40551438, DOI 10.1002/dta.3918, first published 23 June 2025. The abstract and the supplementary table were checked on 20 August 2026. The opened publisher page states no funding details. Its authors declare no conflicts of interest; this disclosure was checked on 5 September 2026.

The most reported molecules were ibutamoren, ligandrol, ostarine, cardarine and andarine (PMID 40551438). Ibutamoren is a growth hormone secretagogue. The authors state that most samples came from illegal distribution. That describes the study sample, not a rate for the wider market.

From limited amount data, the authors inferred that most samples held active doses. They inferred that some held too much. They did not count the number of overdosed samples. These are the authors' inferences from the abstract (PMID 40551438; checked 20 August 2026).

Australia supplied about 45% of the samples (DOI 10.1002/dta.3918, supplementary table 1; checked 20 August 2026). Australia and Canada together supplied about 48% (same DOI and table; checked 20 August 2026).

These rows show which lab sent data. They do not show how common a product is in that country.

What did the Swedish and German examples show?

The Swedish table lists four samples with unknown labelled components (DOI 10.1002/dta.3918, supplementary table 1; checked 20 August 2026).

These are table entries, not national rates. The table lists 14 samples with more than one substance (DOI 10.1002/dta.3918, supplementary table 1; checked 20 August 2026). What a lab finds depends on what it screened for. This paper reports molecules and amounts. It does not report purity, endotoxin or heavy metals.

What did forensic peptide studies find?

A Belgian medicines lab tested falsified peptide drugs. The Janvier full paper names no funder. The authors state that they had no conflict of interest.

All arsenic found was inorganic. Each preparation had at least one trace residual solvent. All calculated maximum solvent levels were below ICH limits. The source is Janvier et al., Talanta, PMID 30029448, DOI 10.1016/j.talanta.2018.06.023. These findings were checked on 26 August 2026.

A separate lab tested products labelled as follistatin. The results below come from Reichel, Gmeiner and Thevis, PMID 31758732, DOI 10.1002/dta.2741. Funding details were not established from the source material reviewed for this page.

Some other products held growth-promoting peptides, including MGF and GHRP-2 (PMID 31758732). Oligomers are groups of molecules linked together. That term describes a finding in these products, not a test of all follistatin on sale.

In another paper, Danish authorities sent unknown preparations for mass-spectrometry tests. The powders were analogues of GHRP-2, GHRP-6, ipamorelin and modified GRF 1-29 (PMID 30136411; checked 20 August 2026).

Each identified change was an extra glycine at the N-terminus (PMID 30136411). That is the amino end of the peptide chain. This paper identifies what was found. It does not test what the added glycine does in a person. The source is Gajda et al., Drug Testing and Analysis 2019, DOI 10.1002/dta.2489. Funding details were not established from the material reviewed for this page.

What do insulin substitution and counterfeit GLP-1 reports show?

Substitution means a different active drug is present. It is not the same finding as low purity. The reports below concern different products from the test-purchase vials above.

Hypoglycaemia means low blood sugar. Hypokalaemia means low blood potassium. Hypothermia means low body temperature. The technical terms are kept in the table so they can be matched to the papers.

The Italian report is in European Journal of Hospital Pharmacy, PMID 41125326, DOI 10.1136/ejhpharm-2025-004656. The Dutch report is in Nederlands Tijdschrift voor Geneeskunde, PMID 42390457. Both were checked on 20 August 2026. Funding details were not established from the material reviewed for this page.

The source is Frontiers in Pharmacology, PMID 42137313, DOI 10.3389/fphar.2026.1805842. Funding details were not established from the material reviewed for this page.

These are spontaneous reports. They are not an incidence study. The report count does not tell us how often an event occurs among all users. See the separate page on counterfeit GLP-1 products for the cases.

What does purity cover?

A purity result depends on what the test looked for. The EMA guideline sets out the impurities a peptide maker must control. It is not a test of grey-market goods. A lab report needs the name of the method to show the reach of its result. That is how a reader can tell which tests were done.

The source is EMA/CHMP/CVMP/QWP/367182/2025, section 4.3.2. It took effect on 1 June 2026 (same identifier; checked 20 August 2026). This is an agency guideline, not a funded trial.

Source for each entry: EMA/CHMP/CVMP/QWP/367182/2025, section 4.3.2, checked 20 August 2026. The first groups concern the peptide itself. The last row lists impurities that are not peptides.

The test method must state what it looked for. A purity number alone does not supply that detail.

What do these measurements not prove?

None of these datasets is a random survey of injectable research-peptide vials across the market. The largest one consists of reports that people chose to submit. The purchase study describes the goods received in that purchase. The official-lab study has a broader scope than peptides.

The Janvier study covers the samples that team tested. The EU database holds reports of suspected events. Each source shows what was measured or reported in its own sample. It does not identify an untested vial's contents. Nor does it turn a certificate, label or product name into proof.

For label claims, read research use only labels. For amounts after water is added, read reconstitution and what a dose measures. For symptoms in published cases, read when to get help. The vial, test and symptom hub links the shared evidence pages.