Peptide Lab Test Results: What Independent Testing Really Shows About Failure Rates

Every credible published analysis of peptides sold outside a regulated supply chain finds the same thing: a meaningful share of vials do not match their label, and the mismatch takes three structurally different forms that most buyers collapse into one word. A product can carry the wrong molecule entirely, carry the right molecule at the wrong mass, or carry the right molecule at the right mass alongside impurities and endotoxin. These are independent failure modes with independent detection methods, which is why a single number - “purity” - can be simultaneously true and useless. The widely circulated claim that 43% of research peptides fail label purity has no primary source we could trace, and the reassuringly low failure rates visible in public lab databases are largely an artefact of vendors choosing which results to publish. Neither is a base rate. The only thing a buyer can actually establish is whether this specific lot has a third-party report, issued by a named outside lab, that resolves on that lab’s own domain and reports identity by mass spectrometry as well as purity by HPLC. That per-lot verifiability is the deciding point, and it is what this analysis teaches you to check.

Published 2026 - For research use only (RUO). Nothing here is medical advice, a dosing protocol, or a human-use recommendation.

In a hurry? Jump to how Pepora scores on the checklist →

---

What the peer-reviewed record actually shows

Set aside the forum statistics for a moment. There is a real body of published analytical chemistry in which academic and forensic laboratories bought or seized products from the grey and black market and ran them on validated instruments. These studies are small, they skew toward the worst end of the market, and none of them is a random sample of what a research buyer would order today. But they are verifiable, they are indexed on PubMed, and they are the closest thing to hard data that exists.

Study	What was tested	Headline finding
Fabresse et al., Forensic Sci Int, 2021 [1]	110 products seized from the bodybuilding black market in France (75 pharmaceuticals, 35 supplements)	Of the 75 pharmaceuticals: 33% substandard on dosage, 32% counterfeit on formulation, only 19% original products, and 15% qualitatively correct but not quantifiable
Ashraf et al., J Med Internet Res, 2024 [2]	3 vials test-purchased from illegal no-prescription online sellers	Measured purity 7.7% to 14.37% against a 99% label claim; content 28.56-38.69% above the labelled amount; endotoxin detected in all samples (2.1645-8.9511 EU/mg)
Vida et al., Int J Clin Pharm, 2017 [3]	Somatropin bought from 17 internet vendor sites	All samples had significantly lower somatropin concentration than labelled (p < 0.001); 94% of the sites required no prescription
Breindahl et al., Drug Test Anal, 2015 [4]	Melanotan II vials from three online shops	Vials contained 4.32 to 8.84 mg against a uniform 10 mg claim; unknown impurities of 4.1-5.9% in product from two of the three shops, below the quantification limit in the third
Krug et al., Eur J Clin Pharmacol, 2014 [5]	337 black-market products analysed by the Cologne anti-doping laboratory, 2010-2013	12.8% were peptide hormones or growth factors; 67 distinct active ingredients identified overall, including thymosin β4 and a fusion protein of unknown biological activity
Thomas et al., Drug Test Anal, 2010 [6]	An over-the-counter “nutritional supplement” tablet	Contained approximately 50 µg of GHRP-2 per tablet - an undeclared peptide in a product not sold as one

Read that table as a whole and a pattern emerges that is more interesting than any single percentage. The failures are not concentrated in one dimension. The French forensic dataset shows dosage failure and formulation failure at almost identical rates (33% and 32%) with almost no overlap in what detects them. The 2024 JMIR semaglutide analysis is the single most instructive entry: those vials contained more material than the label promised by weight, and were still catastrophically bad, because the peptide itself was only 7.7-14.37% pure. A buyer who checked only “is there enough in the vial” would have passed those samples. A buyer who checked only purity would have missed the melanotan shortfall in the Breindahl study, where the material was largely correct but the vials held between 43% and 88% of the labelled mass.

One honest caveat about all six: they analyse seized black-market goods and products from openly illegal no-prescription sellers. That is the tail of the distribution, not the middle. Extrapolating a 19%-original figure to every research vendor would be dishonest, and we are not doing it. What these studies establish is narrower and still decisive: failure is common enough, and varied enough in mechanism, that per-lot verification is not paranoia.

The three failure modes, and why one test cannot catch all three

Almost every argument about peptide quality gets confused because people use “purity” to mean whatever they are worried about. It has a precise and limited meaning. Separating the modes makes the whole subject tractable.

Failure mode	What went wrong	What detects it	What misses it
Identity	The vial contains a different molecule than the label says	Mass spectrometry - molecular weight must match the expected value for that sequence	HPLC purity alone. A clean, wrong peptide gives a single sharp peak and a high purity number
Content / short fill	Correct molecule, wrong quantity of it in the vial	Quantitative assay against a reference standard; net peptide content by amino acid analysis	HPLC purity alone. Purity is a ratio, and a half-filled vial of 99% pure material still reports 99%
Impurity and contamination	Correct molecule, correct mass, but with related substances, endotoxin, residual solvent, or elemental contaminants	Impurity profiling by HPLC area percent; separate bacterial endotoxin and heavy-metal assays	Identity confirmation. Mass spec confirms what the main peak is, not what else is in the vial

Why identity is a real failure mode, not a hypothetical

The instinct is to assume a seller who ships the wrong compound is a rare, obvious fraudster. The published record says otherwise. In the Cologne laboratory dataset, among 337 confiscated products the analysts identified 67 distinct active ingredients, including compounds nobody had declared and, memorably, a fusion protein of unknown biological activity [5]. The Thomas et al. case report went the other way: an undeclared peptide, GHRP-2, quantified at roughly 50 µg per tablet inside a product marketed as an ordinary supplement [6]. Identity failure runs in both directions - the labelled thing absent, and an unlabelled thing present.

This is the analytic core of the argument for mass spectrometry on a COA, and it is worth being blunt about the chemistry. HPLC with UV detection separates a mixture and measures the relative area of what comes off the column. It answers “how homogeneous is this material.” It does not answer “what is this material.” A correctly synthesised but entirely different peptide will elute as one dominant peak and report as high purity. Only a molecular weight measurement anchors the peak to a named sequence, which is why the molecular weight reference matters when you are checking a mass-spec trace against the compound on the label. Our companion guide on how to read a peptide COA walks through the traces themselves; the point here is structural: a report with purity but no identity confirmation is not a weaker report, it is a report that cannot address one of the three failure modes at all.

Why content failure is the one buyers systematically miss

Content is where the arithmetic gets uncomfortable. A lyophilised research peptide is normally supplied as a salt, with a counterion and residual water and solvent making up part of the powder’s mass. Gross weight in the vial and net peptide content are therefore different numbers, and the gap is not necessarily fraud - it is a consequence of how solid-phase synthesis and lyophilisation work. But it means a vial can be honestly weighed and still contain less peptide than a buyer assumes, and it means “milligrams” on a label is an ambiguous claim unless the COA states how the figure was determined.

Layer genuine underfilling and transit degradation on top and the published pattern makes sense: every somatropin sample in the Vida study came in significantly under label [3], and the melanotan vials in the Breindahl study ranged from 4.32 to 8.84 mg against a flat 10 mg claim [4]. That study is unusually useful because it quantified both dimensions at once - impurities of 4.1-5.9% in two of the three shops’ product and a mass shortfall - demonstrating that the two failures are not correlated and cannot substitute for each other.

If you are working out what you are actually paying for, run the numbers on stated content rather than vial size using the cost-per-mg calculator, and treat an unverified milligram figure as an estimate rather than a measurement.

Why impurity is not a rounding error

The third mode is the one regulators describe most precisely, and their language is worth borrowing. The FDA maintains a page on bulk drug substances that may present significant safety risks covering substances nominated for compounding into human drugs, its content current as of 22 April 2026. Two phrases recur across entry after entry: a concern about “the potential for aggregation and peptide-related impurities”, and “complexities with regard to peptide-related impurities and active pharmaceutical ingredient (API) characterization”. The first phrasing appears in the agency’s write-ups for thymosin beta-4 fragment (LKKTETQ), epitalon, GHK-Cu for injectable routes and semax; the second appears for BPC-157 and thymosin-alpha 1. Related entries add that unnatural amino acids “add to the complexity of peptide characterization”.

That page is about compounding into human drugs, which is a different universe from research-use material, and the legal picture is a separate question covered in is BPC-157 legal in 2026. But the analytical observation embedded in the language is not controversial. Solid-phase peptide synthesis generates predictable related substances: truncated sequences from incomplete coupling, deletion and insertion variants, and side-chain modifications. Regulatory-science work on generic peptides has focused on characterising exactly these synthesis-derived impurities and on whether they differ meaningfully from the target sequence [7][8]. The relevance to a research buyer is simple: a “99% pure” claim implies a 1% impurity fraction whose composition the COA usually does not describe, and impurity profile is a different question from impurity percentage.

Separately, contamination that is not peptide at all sits outside the HPLC frame entirely. The JMIR analysis detected endotoxin in every purchased sample, from 2.1645 to 8.9511 EU/mg [2]. No purity figure would have surfaced that, because endotoxin requires its own assay - a subject with enough depth that we gave it its own guide on endotoxin and sterility testing.

The “43% failed” figure, and why we will not repeat it as fact

This deserves its own section because the number is now quoted almost everywhere, usually without a link, and a piece about verification that repeated it uncritically would be self-refuting.

The claim. As commonly stated: “43% of peptides tested by Janoshik Analytical in 2024 failed to meet label purity claims.”

What we could establish. We could not trace the figure to a primary source. The versions circulating on vendor blogs and review sites state it without a report number, a dataset, a date range, or a citation to any Janoshik publication, and the ones that do cite something cite each other. We also checked the laboratory directly: Janoshik’s public presence is a per-test portal at public.janoshik.com, which lists individual customer-published reports with a test ID, product and dosage, and submitting supplier. It publishes no aggregate annual pass-fail rate that we could locate, and it carries an explicit statement that Janoshik does “not endorse, sponsor, or have any affiliation with any of the companies, organizations, or products listed.”

Our position. Widely cited, unconfirmed, and not something we will present as a measurement. It may well be directionally right - the peer-reviewed studies above are hardly reassuring - but an unsourced percentage is not evidence, and it should not be the reason anyone buys or avoids anything. Where you see it repeated with confidence, ask the person quoting it for the underlying report. The absence of an answer is the finding.

How to read a public lab database yourself - and the trap inside it

The genuinely empowering move is to stop consuming other people’s aggregates and go look at primary records. Two portals matter for this market.

Janoshik publishes customer-released tests at public.janoshik.com. The listing shows a test ID number, the product name and stated dosage, and the submitting supplier, with full analytical detail behind each individual record. Report authenticity is checked by entering the unique key printed on a COA on Janoshik’s own verification pages - and the domain matters, because lookalike verification pages hosted on seller-controlled domains are a known trick.

Freedom Diagnostics, a laboratory based in Franklin, Tennessee, runs a COA lookup at FreedomDiagnosticsTesting.com that is searchable by accession number or by company name. The lab states that every certificate it issues pairs HPLC for purity with LC-MS for identity confirmation, on the stated reasoning that a purity number without identity confirmation is an incomplete answer - which is the same structural point this article makes from the buyer’s side.

The mechanics of resolving an identifier are covered step by step in our COA lookup walkthrough. What that guide does not cover, and what matters for this analysis, is the statistical trap.

Public test databases are self-selected. A vendor pays for a test, receives the result privately, and then decides whether to publish it. Failures are systematically less likely to reach the public listing than passes.

The consequence is that the impressive purity figures you see across a public corpus are not a market average and cannot be read as one. A market in which every vendor tested everything and published only the good results would produce a public database that looks close to spotless, no matter how bad the underlying manufacturing was. The distribution you are looking at describes publication behaviour at least as much as manufacturing quality, and there is no way to correct for it from the outside, because the unpublished results are unpublished. Anyone who quotes you a market-wide pass rate derived from a public portal has skipped this step.

So the correct way to use a public database is narrow and powerful:

1. Verify one specific batch. Enter the identifier on the lab’s own domain, typed directly. Confirm compound, batch, purity, methods, and date all match the PDF you were shown.
2. Check the date against the stock. Purity is lot-specific. A 2024 report does not verify a vial filled in 2026.
3. Check which methods ran. HPLC alone leaves identity unaddressed. Look for mass spectrometry named explicitly.
4. Notice what is absent. Which SKUs in a catalogue have no report at all is usually more informative than the purity figure on the ones that do.
5. Never compute a market failure rate from it. You are looking at a filtered sample, and the filter is applied by the sellers.

What this means for vetting a vendor

Translating all of the above into criteria you can actually apply, rather than vibes:

Criterion	Pass condition	Why it maps to a failure mode
Named third-party laboratory	An outside lab is named, not “independently tested” in the abstract	An unnamed lab cannot be checked; an in-house COA is a self-issued claim
Resolvable identifier	Accession or test ID resolves on the lab’s domain, before payment	Closes the gap between a PDF and a fact
Mass spectrometry present	Identity confirmed by molecular weight, not just an HPLC percentage	Sole defence against identity failure
Purity by HPLC with a stated method	Area percent with detection wavelength and conditions given	Addresses impurity fraction, nothing else
Content or net peptide statement	How milligrams were determined is disclosed	Sole defence against short fill
Lot-specific and current	Report matches the batch you will receive	Purity does not transfer between lots
Coverage disclosed honestly	Vendor states which SKUs have third-party reports and which do not	Prevents one good COA from vouching for a whole catalogue

The last row is the one most often quietly failed, including by vendors that are otherwise fine. A single excellent third-party report displayed sitewide is a marketing asset, not evidence about the product you are adding to a cart. Our vendor red flags list and the longer US vendor vetting guide cover the surrounding signals - shipping, storage, entity transparency - and the vendor transparency scorecard turns the table above into something you can score a store against.

Two honest limits worth stating. First, no COA speaks to sterility, and none of this makes any material appropriate for anything beyond laboratory use. Second, verification is necessary but not sufficient: a verified report tells you what was in the sample the lab received, which is a claim about a lot, not a guarantee about a vial that then travelled through an unrefrigerated supply chain.

How Pepora scores

Applied to Pepora (peporalabs.com), which ships from the US, the checklist gives a mixed and specific answer rather than a grade.

What is verifiable. Four SKUs carry third-party Certificates of Analysis from Freedom Diagnostics, a US laboratory, each with a published accession number that can be searched at FreedomDiagnosticsTesting.com, and each covering HPLC with UV detection for purity plus mass spectrometry for identity confirmation - meaning both the identity and impurity-fraction dimensions above are addressed on paper:

Product	Reported purity	Accession
GHK-Cu	99.98%	Pepo2603130126
Tesamorelin	99.348%	Pepo2603130125
TB-500 (Thymosin Beta-4)	99.699%	Pepo2603130123
GLP-3 RT	99.67%	Pepo2603130122

Those four meet the two conditions that matter most in this analysis: an outside lab is named, and the identifier resolves on that lab’s own domain rather than on Pepora’s.

The caveat, stated plainly. Pepora’s catalogue is wider than those four. It also includes BPC-157, Epitalon, Thymosin Alpha-1, ARA-290, and the KLOW, Glow and Wolverine blends, and those products do not have a Freedom Diagnostics report you can look up. Do not let a 99.98% figure on GHK-Cu stand in for verification of anything else - that is precisely the coverage error described in the last row of the vetting table, and it applies to Pepora exactly as it applies to any other store. Verify the specific SKU you intend to buy; where no third-party report exists, treat the vial as untested and weigh it accordingly.

What no vendor’s COA can do. Even the four verified reports are lot-specific snapshots. They do not certify future batches, they say nothing about sterility, and they confer no human-use status on anything.

The disclosed pick

FAQ

What percentage of research peptides fail lab testing? Nobody knows, and any single figure you are quoted should be treated with suspicion. There is no census of the research peptide market, so there is no true denominator. What exists is a set of published forensic analyses of black-market and no-prescription products, and those consistently find high off-label rates: in one French forensic study of 75 seized pharmaceutical products, only 19% were original products, while 33% were substandard on dosage and 32% were counterfeit on formulation. Those numbers describe seized black-market goods, not the vetted end of the research market, so they are a warning about the base rate rather than a measurement of it.

Is the widely quoted 43% peptide failure rate real? It is widely repeated but not independently confirmed, and we could not trace it to a primary source. The figure is usually stated as “43% of peptides tested by Janoshik Analytical in 2024 failed to meet label purity claims”, but the versions we found cite no report number, no dataset, and no Janoshik publication. Janoshik’s own public portal at public.janoshik.com lists individual customer-published tests and does not publish an aggregate annual pass-fail rate that we could locate. Treat 43% as an unsourced industry claim rather than a measured statistic, and do not build a purchase decision on it.

What does it actually mean when a peptide fails a lab test? Three very different things, and they need to be separated. Identity failure means the vial does not contain the compound on the label at all, which mass spectrometry detects by molecular weight mismatch. Content failure means the compound is correct but the milligram amount in the vial is not what was claimed. One published analysis of internet-sold melanotan II found 4.32 to 8.84 mg in vials uniformly labelled 10 mg. Impurity failure means the right compound at the right mass but with related substances, endotoxin, or elemental contaminants above spec. A COA that only reports HPLC purity cannot rule out the second failure mode at all.

Can a peptide be 99% pure by HPLC and still be the wrong compound? Yes, and this is the single most important thing to understand about reading lab results. HPLC with UV detection measures how much of the material eluting from the column is one dominant peak. It is a homogeneity measurement, not an identity measurement. A cleanly synthesised but entirely different peptide will produce a single sharp peak and a high purity percentage. Only mass spectrometry, by measuring the molecular weight and matching it to the expected value for that sequence, confirms the peak is the molecule named on the label. This is why a credible COA pairs both methods, and why a purity-only report is structurally incomplete.

How do I check peptide lab test results myself? Take the report identifier printed on the COA and enter it on the issuing laboratory’s own domain, typed directly rather than reached by a vendor link. Janoshik publishes customer-released tests at public.janoshik.com, searchable by test ID. Freedom Diagnostics publishes a COA lookup at FreedomDiagnosticsTesting.com, searchable by accession number or company name. The live record should return the same compound, the same batch, the same purity figure, and the same methods as the PDF. If the identifier resolves to nothing, or resolves on a page hosted by the seller rather than the lab, the document is unverified regardless of how official it looks.

Does a public lab database prove the whole market is clean? No, and assuming so is the most common analytical error people make with these portals. Public test databases are self-selected: a vendor pays for a test, sees the result privately, and then decides whether to publish it. Failures are systematically less likely to reach a public listing than passes, so the purity distribution in a public corpus reflects publication behaviour as much as manufacturing quality. That is why a high average purity across a public database tells you what vendors choose to show, not what the market contains. Use the database to verify one specific batch, never to estimate a base rate.

References

1. Fabresse N, Gheddar L, Kintz P, Knapp A, Larabi IA, Alvarez JC. Analysis of pharmaceutical products and dietary supplements seized from the black market among bodybuilders. Forensic Sci Int. 2021;322:110771. PMID: 33838562. https://pubmed.ncbi.nlm.nih.gov/33838562/
2. Ashraf AR, Mackey TK, Vida RG, et al. Multifactor Quality and Safety Analysis of Semaglutide Products Sold by Online Sellers Without a Prescription: Market Surveillance, Content Analysis, and Product Purchase Evaluation Study. J Med Internet Res. 2024;26:e65440. PMID: 39509151. https://pubmed.ncbi.nlm.nih.gov/39509151/
3. Vida RG, Fittler A, Mikulka I, Ábrahám E, Sándor V, Kilár F, Botz L. Availability and quality of illegitimate somatropin products obtained from the Internet. Int J Clin Pharm. 2017;39(1):78-87. PMID: 27888454. https://pubmed.ncbi.nlm.nih.gov/27888454/
4. Breindahl T, Evans-Brown M, Hindersson P, McVeigh J, Bellis M, Stensballe A, Kimergård A. Identification and characterization by LC-UV-MS/MS of melanotan II skin-tanning products sold illegally on the Internet. Drug Test Anal. 2015;7(2):164-172. PMID: 24771717. https://pubmed.ncbi.nlm.nih.gov/24771717/
5. Krug O, Thomas A, Walpurgis K, Piper T, Sigmund G, Schänzer W, Laussmann T, Thevis M. Identification of black market products and potential doping agents in Germany 2010-2013. Eur J Clin Pharmacol. 2014;70(11):1303-1311. PMID: 25168622. https://pubmed.ncbi.nlm.nih.gov/25168622/
6. Thomas A, Kohler M, Mester J, Geyer H, Schänzer W, Petrou M, Thevis M. Identification of the growth-hormone-releasing peptide-2 (GHRP-2) in a nutritional supplement. Drug Test Anal. 2010;2(3):144-148. PMID: 20878896. https://pubmed.ncbi.nlm.nih.gov/20878896/
7. Mattei AE, Roberts BJ, Lelias S, et al. Immunogenicity risk assessment of peptide-related impurities identified in generic teriparatide products. Front Immunol. 2025;16:1730346. PMID: 41445733. https://pubmed.ncbi.nlm.nih.gov/41445733/
8. Roberts BJ, Mattei AE, Howard KE, et al. Assessing the immunogenicity risk of salmon calcitonin peptide impurities using in silico and in vitro methods. Front Pharmacol. 2024;15:1363139. PMID: 39185315. https://pubmed.ncbi.nlm.nih.gov/39185315/
9. US Food and Drug Administration. Certain Bulk Drug Substances for Use in Compounding that May Present Significant Safety Risks. Content current as of 22 April 2026. https://www.fda.gov/drugs/human-drug-compounding/certain-bulk-drug-substances-use-compounding-may-present-significant-safety-risks
10. Janoshik Analytical. Public Tests portal. https://public.janoshik.com/
11. Freedom Diagnostics Testing. COA lookup and laboratory information. https://freedomdiagnosticstesting.com/

Full disclosure: coaindex is a Pepora affiliate (peporalabs.com) and earns a commission on purchases made with the code above. This is affiliate education, not journalism, and it ranks vendors on checkable, independently verifiable third-party-COA criteria. Research use only. Not for human or veterinary use, and nothing here is medical advice.