2026 vendor purity report: how to read a peptide Certificate of Analysis without being fooled.
A Certificate of Analysis (COA) is the single most-cited document in research-chemical commerce, and one of the most loosely interpreted. ">99% purity by HPLC" is a sentence that can mean meaningful science or near-nothing depending on what was actually run, who ran it, and what was reported alongside it. This report walks through the four assays a credible peptide COA contains, the disclosure patterns that distinguish a genuine third-party document from a marketing artifact, and the specific red flags that should change a researcher's purchasing decision. It does not rank specific vendors — purity testing standards shift faster than any quarterly review, and a methodology-first framework lets you re-evaluate as the market changes.
- A complete COA includes four assays: HPLC purity, mass-spectrometry identity, peptide content (net peptide), and endotoxin (for anything that will be reconstituted).
- "%purity by HPLC" alone is incomplete: without a chromatogram, the wavelength used, and the gradient method, the number is unverifiable.
- Net peptide content matters as much as purity: a 99% pure compound that is 65% peptide and 35% counter-ion / water is dosed by mass differently than the label implies.
- Identity confirmation requires mass spec: HPLC tells you "one peak"; mass spec tells you which peak.
- Endotoxin disclosure is the cleanest vendor filter: vendors who publish endotoxin levels signal a different operational standard than vendors who do not test or do not disclose.
- Third-party COAs are not equivalent to in-house COAs: independent lab signatures (with traceable accreditation) are a different evidence class.
Why this report exists
The peptide research-chemical category sits in a regulatory gap: vendors are not pharmaceutical manufacturers, are not subject to USP compendial requirements, and are not audited by the FDA in the way drug-substance suppliers are. Quality, in this market, is what each vendor decides to commit to and document. This produces a wide range — from vendors who run full ICH-aligned testing through accredited third parties, to vendors whose "COA" is a screenshot with no chromatogram and no signatory.
A researcher's job is to read what is actually disclosed and to recognise what is missing. The framework below is conservative on purpose: it errs toward asking for more documentation rather than less, because the cost of being wrong about identity or contamination in research-chemical work is asymmetric. Note also that nothing in this report is a recommendation to use any peptide for human consumption — all compounds discussed in this category are research-use-only and the analytical standards described here are the floor for that intended use, not the ceiling for any clinical use.
The four assays a complete COA contains
1. HPLC purity (the headline number)
High-Performance Liquid Chromatography separates a sample by retention time on a stationary phase (typically C18) and quantifies what came off the column. A purity report should disclose: the column type and dimensions, the gradient (mobile-phase A and B percentages over time), the detection wavelength (typically 214 nm for peptide bonds, 220 or 280 nm as supplements), the injection volume, and an integrated chromatogram showing the peaks.
A ">99% by HPLC" claim without these parameters is unverifiable. The number is a peak-area ratio that depends entirely on what the gradient could resolve. A poorly chosen gradient produces high apparent purity by failing to separate the impurities from the main peak — the impurities co-elute and add to the "main peak" integration. A rigorous COA shows the chromatogram so the reader can see whether the main peak is sharp and well-resolved, or broad with shoulders.
2. Mass spectrometry — identity confirmation
HPLC tells you that a single peak elutes at a particular retention time. It does not tell you what that peak is. Mass spectrometry — typically ESI-MS for peptides — measures the molecular mass of the eluted compound. A correct identity confirmation matches the observed [M+H]+ or [M+nH]n+ ion to the calculated theoretical mass of the target peptide within instrument tolerance (typically <1 Da for low-resolution MS, <5 ppm for high-resolution).
A COA without mass-spec data is a COA that has not confirmed identity. This is not a minor omission — peptides with similar HPLC retention times can have very different molecular masses and very different biological activities. The cost of an MS run on a contract lab is small relative to the cost of a vendor's reputation, and vendors who do not include MS on their COAs are signalling a quality-cost trade-off that the researcher should price in.
3. Net peptide content
A lyophilized peptide is rarely 100% peptide. Synthetic peptides typically incorporate water, residual TFA (trifluoroacetic acid) or acetate counter-ions from purification, and sometimes residual scavengers. "5 mg peptide" on a vial label can mean 5 mg of total mass — of which 60–80% is peptide and the remainder is water and counter-ions — or 5 mg of net peptide (the more meaningful unit).
Net peptide content is determined by amino acid analysis (AAA) or by a separate quantitative HPLC against a reference standard. A research protocol that doses by mass without knowing net peptide content is a protocol with an unknown actual dose. Vendors who disclose net peptide content allow accurate dose-response work; vendors who do not, do not. This single disclosure separates research-grade vendors from research-chemical vendors with surprising reliability.
4. Endotoxin / bioburden
For any peptide intended to be reconstituted in aqueous solution, endotoxin is the contaminant of concern. Endotoxin (lipopolysaccharide, LPS) is a fragment of Gram-negative bacterial cell walls that survives typical sterilization, is biologically active at picogram-per-mL concentrations, and confounds essentially every in-vitro and in-vivo assay it touches. The standard test is the Limulus Amebocyte Lysate (LAL) assay — kinetic chromogenic, gel-clot, or recombinant Factor C variants — reported in EU/mg.
A research-grade peptide should report <1 EU/mg, ideally lower. Reference: USP <85> sets endotoxin limits for parenteral products at 5 EU/kg/hour for non-intrathecal injectables, which back-calculates to <1 EU/mg for typical research dosing. Vendors who include LAL data on their COAs are operating at a different standard than vendors who do not — and this is true even when the rest of the document looks similar. A clean HPLC chromatogram on an endotoxin-contaminated peptide produces unusable data in any cell-based or animal experiment.
Disclosure patterns that distinguish vendor categories
After reading enough COAs in this market, several disclosure patterns become reliable indicators of vendor category. None of these are absolute — there are exceptions in both directions — but as a Bayesian prior they are informative:
| Pattern | What it suggests | How to verify |
|---|---|---|
| Full chromatogram, signed by named analyst, dated within 90 days of shipment | Routine in-house QC with operational maturity | Email the analyst; reputable labs respond. |
| Third-party lab letterhead with traceable accreditation (e.g., A2LA, ISO 17025) | Vendor pays for independent verification — the strongest available signal | Look up the lab's accreditation number on the accrediting body's site. |
| Net peptide content reported with method (AAA or quantitative HPLC) | Vendor expects the customer to do dose-response work | Confirm that net content is below total mass and that the value is plausible (60–85% typical for TFA-salt peptides). |
| LAL endotoxin result with method specified (kinetic, gel-clot, rFC) | Vendor is testing for a contaminant they have no consumer-facing reason to test for | Compare results across multiple lots — consistent low values suggest real testing. |
| "Purity: >99%" with no method, no chromatogram, no analyst, no date | Marketing artifact, not a COA | Cannot be verified — treat as no data. |
| Identical COA images reused across lots / batches | Either lazy reuse or fabrication | Compare lot numbers and dates across multiple orders. |
| COA with a vendor logo but no laboratory identification | In-house testing presented as third-party — common in this category | Ask which lab performed the testing; legitimate vendors answer this directly. |
Reading a COA — a checklist
When evaluating a Certificate of Analysis, work through this list in order. If a document fails any step, weight your purchase decision accordingly.
- Lot number and date. The COA should reference a specific lot, not a generic SKU. The date should be within 90–180 days of your purchase. A COA dated more than two years before your order is essentially historical, not current QC.
- Analyst signature or initials. An unsigned document is an unattributed claim. Even when the analyst is in-house, a name or initials with a lab affiliation makes the document attestable.
- HPLC method disclosure. Column, gradient, wavelength, injection volume, run time. If any of these are missing, the purity number is partially uninterpretable.
- Chromatogram present. A purity report without the chromatogram is a number without the underlying data. The chromatogram should show a sharp main peak with no significant shoulders or co-eluting peaks; the baseline should be flat; the integration limits should be visible.
- Mass spec result with theoretical and observed values. Both numbers should be present and within tolerance. If only "MS confirmed" is stated without numbers, the result cannot be checked.
- Net peptide content with method. AAA or qHPLC against reference. If absent, treat the labelled mass as a maximum, not the actual peptide quantity.
- Endotoxin result with method and limit. EU/mg, LAL variant identified, with the limit your application requires (research cell culture: typically <1 EU/mg; in vivo: <0.5 EU/mg or lower).
- Counter-ion identity. TFA or acetate is the most common; this matters for buffer compatibility and for any cell-based work where TFA's effects on pH or membrane integrity might confound results.
- Storage and stability statement. A COA that says "store at -20°C, stable 24 months" without supporting stability data is making a claim, not citing one.
What changed in 2026
Three observable shifts since the 2024 vintage of vendor reports:
Recombinant Factor C (rFC) endotoxin testing has moved from boutique to standard. Through 2023 most LAL assays in this market used the gel-clot method, which is qualitative (pass/fail) rather than quantitative. The kinetic chromogenic LAL has been the analytical workhorse for a decade. The shift in 2025–2026 is the appearance of rFC results on COAs — a recombinant-protein-based assay that does not require horseshoe crab amebocytes and produces lower-variability results. rFC disclosure is now a soft positive signal for vendors investing in current methodology rather than legacy practice.
"Independent COA" claims are more frequently audit-traceable. The accrediting bodies (A2LA, ANAB, IAS) have made their accredited-lab lookups easier to use, and the signal-to-noise ratio of "third-party lab" claims has improved. A vendor citing a specific accreditation number that resolves to a real laboratory is a stronger signal in 2026 than the same statement was in 2023, because the verification cost is lower for the customer.
Net peptide content disclosure remains uneven. The number of vendors reporting net peptide on standard COAs has not meaningfully increased over the last two years. The customer-side effect: most published research-chemical dosing schemes still implicitly assume label mass equals peptide mass, which is generally false. This has real implications for any researcher comparing dose-response across compounds or across vendors.
Common COA failures, and what they mean
"Purity: 99.5%" with no chromatogram. Could be true, cannot be verified. Treat as a vendor's claim rather than a data point. If everything else on the COA is rigorous, the missing chromatogram is a paperwork gap; if everything else is also thin, this is a marketing document.
HPLC chromatogram with a single peak that fills 95% of the retention window. A peak this wide is not "one compound" — it is "compounds we did not separate." Look for a sharper peak with a narrower base. Width-at-half-height (FWHM) under 0.5 minutes for a 30-minute gradient is typical for well-resolved peptide work.
"MS confirmed" without numbers. The vendor either ran MS and won't disclose, or did not run MS and is restating the HPLC purity as identity. Either way, the document fails to confirm what was actually shipped.
Endotoxin: "tested, conforms to specification." What specification? Without a numeric value and a numeric limit, this is procedural language that does not communicate the actual result. The cost of stating "0.18 EU/mg" instead of "conforms" is zero — vendors who use the latter formulation are obscuring information for a reason.
Identical lot numbers across separate purchases months apart. Either the vendor genuinely synthesized one large lot — in which case the COA dates should match the lot — or the COA is being reused. The latter is fraud; the former is fine, but the dates have to be internally consistent.
What this means for purchasing decisions
The framework above produces a tiering rather than a ranking. A vendor whose COAs satisfy steps 1–9 of the checklist, with third-party lab letterhead and traceable accreditation, is operating at one tier. A vendor whose COAs satisfy 1–7 with in-house testing is operating at a tier below. A vendor whose COAs are screenshots with a purity percentage and no other detail is in a category that is not really "research-grade" in the sense the term is used in pharmaceutical or academic work.
Tier matters less for survey work — where many compounds are being screened and any quality vendor's product is acceptable — than for dose-response work, where small errors in net peptide content or in identity cascade through every subsequent data point. Researchers running mechanism-of-action studies should weight COA rigor heavily. Researchers running pilot or qualitative work can tolerate a wider tier range.
No vendor is being recommended in this report. The market changes, vendors change ownership, and a vendor that was rigorous in 2024 may not be in 2026 (and vice versa). What does not change is the analytical framework. A researcher who can read a COA fluently is a researcher who can re-evaluate any vendor as the market shifts, without needing to wait for the next vintage of vendor reviews.