How to Read a Peptide Certificate of Analysis

Updated August 2026 · Patriot Labs Research Team · ~26 min read

What a peptide COA is

A peptide Certificate of Analysis (COA) is a batch-specific laboratory report stating what one lot of peptide was tested for, which methods were used, and what results those methods returned. In practice it answers two questions — is this the compound it claims to be (identity) and how much of what is in the vial is that compound (purity, and separately content) — and it is valid only for the single lot number printed on it.

That is the whole idea. Everything else in this guide is detail about how those two questions get answered, how the answers can be technically true and still misleading, and how to tell a report produced by an instrument from a graphic produced in a design tool.

The word certificate oversells it. A COA certifies nothing about the material in a general sense. It certifies that a named laboratory ran named tests on a named lot on a named date and got the printed numbers. Every one of those four nouns is load-bearing, and a COA missing any of them has lost most of its meaning. That is the first practical lesson: a COA is only as good as its traceability, and traceability is the part that is cheapest to fake and easiest to check.

It is worth being blunt about why this matters. A 2024 market-surveillance study bought semaglutide products from online sellers operating without prescription requirements. Of six purchases, three were never delivered. The three vials that did arrive measured 7.7% to 14.37% purity by liquid chromatography–mass spectrometry against vendor claims of 99%, carried detectable endotoxin, and were classified as substandard and falsified.^[28] The gap between a claimed number and a measured number can be an order of magnitude, and the claimed number is the one printed on the marketing.

Anatomy of a COA, field by field

Real certificates vary in layout, but the information set is fairly standard. Here is what each field is for and what a careful reader checks.

That last row is not a stylistic preference. In the United States, the regulation governing pharmaceutical laboratory records requires that records include a description of the sample with its source, quantity, lot number, and the dates it was taken and received; the analytical method used; all calculations; a statement of results and how they compare with established standards of identity, strength, quality and purity; and the initials or signature of the person who performed the test plus a second person who checked it.^[29] Research-grade material is not sold under those rules, but the list is a useful yardstick for what a serious analytical record looks like. When a COA carries five of those elements instead of one, that tells you something about the culture that produced it.

Purity by RP-HPLC: how the number is made

Nearly every purity figure on a peptide COA comes from reversed-phase high-performance liquid chromatography (RP-HPLC). Understanding how that number is generated is the difference between reading a COA and being shown one.

A small amount of the peptide is dissolved and injected onto a column packed with silica particles whose surfaces are covalently coated in hydrocarbon chains — most commonly C18, an eighteen-carbon chain, which is the default stationary phase for peptide work, with shorter C8 or C4 phases used for very hydrophobic sequences.^[9] That surface is non-polar. The mobile phase pumped through it starts polar — mostly water — and becomes progressively less polar as acetonitrile is blended in over the course of the run. This ramp is the gradient.

Peptides stick to the C18 surface in proportion to how hydrophobic they are, and release when the acetonitrile fraction reaches the point that makes staying on the surface less favourable than travelling in the solvent. Retention of a peptide falls off very sharply with organic content, which is exactly why gradient elution rather than a fixed solvent mixture is used: a single isocratic composition that elutes one peptide will strand another on the column indefinitely.^[9]

A small amount of acid — typically trifluoroacetic acid (TFA) at low concentration — is added to both solvents. TFA does two jobs. It keeps the mobile phase acidic, which suppresses ionisation of residual silanol groups on the silica and stops peptides smearing across the column. And it acts as an ion-pairing agent: the trifluoroacetate anion pairs with protonated basic side chains (lysine, arginine, histidine) and the free N-terminus, masking their charge and sharpening the peaks. The concentration of TFA measurably affects both peak shape and recovery.^[9] Remember this detail — it comes back later, because the TFA that improves the chromatography is the same TFA that ends up in the bottle.

As material comes off the end of the column it passes through a UV detector. The output is a chromatogram: time on the x-axis, absorbance on the y-axis, and a series of peaks. Software integrates the area under each peak. The purity figure is then:

purity % = (area of main peak ÷ total area of all integrated peaks) × 100

This is called area normalisation, and it is worth naming the four assumptions baked into it, because none of them is exactly true.

One: everything absorbs equally. Area normalisation treats one unit of peak area as one unit of substance regardless of what the substance is. It does not. Response depends on the molar extinction coefficient of the species at the detection wavelength, and those differ between peptides — sometimes considerably. The next section is entirely about this.

Two: everything elutes. Anything that irreversibly binds the column, or that is so hydrophilic it comes out in the void volume with the injection front, never gets integrated. It is not in the denominator, so it cannot lower the purity figure.

Three: everything separates. Any impurity that co-elutes with the main peak is counted as main peak. This is not a hypothetical: the impurities hardest to remove in peptide manufacture are precisely those differing from the target by a single residue or a single functional group, which is also exactly the population most likely to sit under the main peak rather than beside it.^[9]

Four: everything absorbs at all. Counterions, water and most residual solvents produce no meaningful UV signal in this region. They are invisible to the method, which means the purity number says nothing about them whatsoever.

None of this makes RP-HPLC purity a bad number. It is the correct answer to how much of the peptide-related material here is the target peptide. It is simply not the answer to the question most buyers think they are asking, which is how much of this powder is the compound I paid for.

Why 214 nm and 280 nm give different answers

Two wavelengths dominate peptide UV detection, and the choice between them changes the purity number. If a COA does not state which was used, part of the figure is unreadable.

214 nm (or 210–220 nm) detects the peptide backbone. The amide bond itself absorbs in this region. Because every peptide bond in the chain contributes, essentially every peptide-related species in the sample is visible, whatever its side chains. This is the standard choice for purity work and the reason pharmacopoeial guidance on peptide purity points at the 210–220 nm window.^[32]

280 nm detects aromatic side chains — principally tryptophan and tyrosine, with a small contribution from cystine. If a peptide contains none of those residues, it is effectively invisible at 280 nm. If it contains one tryptophan, it lights up.

The quantitative picture at 214 nm is well characterised. A study measuring the contributions of individual residues found the peptide bond has a molar extinction coefficient of about 923 M⁻¹ cm⁻¹ at 214 nm; tryptophan absorbs roughly thirty times more than a single peptide bond; phenylalanine, tyrosine and histidine roughly six times more; methionine about the same as a peptide bond; proline, negligible as a free amino acid but roughly three times a peptide bond when incorporated into a chain; and the remaining residues considerably less.^[10]

Follow the consequence through. A truncation impurity that has lost the single tryptophan from the sequence carries a much smaller response factor than the parent. At 214 nm it will be under-represented relative to its true molar abundance — and at 280 nm it may not appear at all. The purity figure rises, and nothing dishonest has occurred. Conversely, a dimer or an aggregate carrying twice the aromatic content is over-represented, and purity falls.

The practical reading rule: a purity figure quoted at 280 nm for a peptide containing aromatic residues will usually look better than the same sample at 214 nm, because a whole class of impurities has been made invisible. A COA that reports purity at 214 nm is being more conservative than one that reports it at 280 nm, and a COA that does not say is not letting you tell.

Two further method parameters deserve the same scrutiny. A steep, fast gradient compresses everything toward the main peak and merges partially resolved impurities into it; a shallow gradient reports lower purity on the identical sample. A short run time never brings off late-eluting hydrophobic impurities, and dimers are often exactly that. A twelve-minute method and a forty-minute method on the same vial are not measuring the same thing. Pharmacopoeial chromatography practice addresses this class of concern through system-suitability requirements including resolution and peak symmetry.^[11]

Purity is not peptide content

This is the most valuable section on this page, and it is the section almost no retail COA addresses.

Purity is chromatographic. It answers: of the peptide-related material the detector saw, what fraction was the target? Net peptide content is gravimetric. It answers: of the physical powder in the vial, what fraction is peptide at all?

These are different questions with different answers, and the second is routinely much lower than the first. Peptide content is defined as the percentage of total peptide relative to everything else present, and it is arrived at by mass balance — subtracting water, counterions and non-peptide impurities from 100%.^[1] A vendor grading system built on purity alone — the familiar tiers of >95%, 90–95%, >80% and immunograde >65% — describes chromatographic purity and says nothing about content.^[32]

Where does the missing mass go? Three places.

Counterion salt. Peptides purified on preparative RP-HPLC with TFA in the mobile phase are isolated as trifluoroacetate salts, with one trifluoroacetate associated with each protonatable basic site.^[20] Trifluoroacetate is not a trace contaminant. A 2025 study quantifying counterions in lyophilised synthetic peptide salts by HPLC-ELSD, ¹⁹F-NMR and FT-IR found trifluoroacetate reaching up to 35% of total weight, and chloride up to 10% after exchange; for one model peptide the measured burden was 0.333 ± 0.008 mg trifluoroacetate per mg of peptide salt as isolated.^[19]

Water. Lyophilised peptide is an amorphous, high-surface-area solid and it is hygroscopic. It picks up atmospheric moisture whenever a vial is opened in ambient air, and it retains bound water that freeze-drying does not remove. In dried biological products generally, the regulatory expectation has historically been that residual moisture should not exceed about 1.0% for most products, with some specific products allowed up to 2.0%.^[24] Research-grade lyophilised peptide is not held to that specification and commonly sits higher.

Residual solvent. Acetonitrile and other process solvents survive lyophilisation at low levels and contribute mass without contributing peptide.

Here is the arithmetic, laid out as an illustrative mass balance. These figures are a teaching construction, not measurements from any particular lot:

Both numbers in that table are honest. A COA reporting only the first is not lying; it is answering a narrower question than the reader assumes. The difference between 98.7% and 79% is the entire reason this section exists, and it is why a COA that reports net peptide content is showing you something a COA that reports only purity is not.

How is content actually determined? Four methods are recognised: quantitative amino acid analysis, nitrogen determination by elemental analysis, UV spectroscopy against a known extinction coefficient, and quantitative NMR.^[1]

Amino acid analysis is the reference method and has real limits. The peptide is hydrolysed — classically in 6 N hydrochloric acid containing 0.1–1.0% phenol, at 110 °C for 24 hours under vacuum or inert atmosphere — and the liberated amino acids are separated and quantified against standards. The pharmacopoeial chapter is candid that hydrolysis is destructive in a residue-dependent way: tryptophan is destroyed, serine and threonine partially destroyed, methionine can oxidise, cysteine is recovered as cystine with poor efficiency, and bonds involving isoleucine and valine cleave incompletely — restricting reliable quantitation to about seventeen amino acids.^[18]

Nitrogen determination is elegantly simple — peptide is the dominant nitrogen-bearing species in the vial, so total nitrogen scales with peptide mass — but it is confounded by any other nitrogen-containing component. UV is fast but requires an aromatic residue and a trustworthy extinction coefficient. Quantitative NMR is increasingly used and requires no compound-specific reference standard, which is its main attraction.

The honest summary: net peptide content is harder, slower and more expensive to determine than chromatographic purity, which is exactly why most COAs omit it. Its presence on a certificate is a meaningful quality signal. Its absence is normal and not by itself damning — but it means the reader should not treat the purity figure as a mass fraction.

Counterion, water and residual solvent

Having established that these three account for the gap, it is worth understanding each properly, because each has consequences beyond arithmetic.

Counterion

A peptide with basic residues carries positive charge at the acidic pH of a preparative RP-HPLC run, and that charge is balanced by an anion. Because TFA is the standard acid for both cleavage from the resin and mobile-phase modification in purification, cationic peptides come off the process as trifluoroacetate salts.^[20] Three counterions dominate peptide pharmaceuticals — trifluoroacetate, acetate and chloride — and most approved peptide drugs are formulated as acetate salts, which requires a deliberate exchange step.^[20]

Why bother exchanging? Because trifluoroacetate is not inert. It can interfere with physicochemical characterisation, and it has documented effects on mammalian cells in culture, including inhibition of proliferation in some systems, with direction and magnitude varying by cell type.^[20] For anyone running a cell-based in-vitro assay, an unquantified trifluoroacetate load is an uncontrolled variable in the tube. Exchange is achievable — 10 mM hydrochloric acid reduced trifluoroacetate below the limit of quantitation of three analytical techniques after one cycle in a 2025 study, which also found counterion identity modulating passive membrane permeation of cell-penetrating peptides.^[19] The salt form is a property of the material, not a footnote.

Quantifying it is a solved problem. Ion chromatography and capillary electrophoresis are the established analytical routes,^[20] and there are dedicated pharmacopoeial procedures for both major counterions: a general chapter for acetic acid in peptides^[22] and a separate one for trifluoroacetic acid in peptides.^[21] The existence of a standalone official chapter for TFA determination should tell you how routine and how consequential this measurement is considered to be in regulated peptide manufacture. Its near-total absence from retail COAs is a gap, not a sign that it does not matter.

Water

Water content is normally determined by Karl Fischer titration, in which iodine reacts stoichiometrically with water and the endpoint is detected electrochemically. The relevant pharmacopoeial chapter covers three approaches — titrimetric, azeotropic and gravimetric — with the titrimetric method the default unless a monograph says otherwise.^[23] The coulometric variant of Karl Fischer needs only around 10 mg of sample, which matters when the entire contents of a vial is small; a gravimetric loss-on-drying determination needs roughly 200 mg, and thermogravimetric analysis can work from about 2 mg.^[24]

The methods do not measure identical things, which is why the method name matters. Loss on drying measures surface moisture and loosely bound water of hydration; Karl Fischer measures surface moisture and most types of bound water, and consequently often reads higher on the same material.^[24] A COA quoting “water content: 4.2%” with no method named is quoting a number whose meaning depends on an undisclosed choice.

Water is not only a mass-balance issue. It is a stability issue. Residual moisture in a lyophilised solid is the medium in which hydrolysis and deamidation proceed, and freeze-dried products are protected by being dry — though the same guidance notes that over-drying can itself be damaging by stripping protective water from around active sites.^[24] This is the analytical foundation under the practical advice in our guides on storing research peptides and how peptides degrade: keeping a vial cold and sealed is not fussiness, it is keeping the water number where the COA found it.

Residual solvent

Acetonitrile from the purification gradient, ether from precipitation steps, and traces from cleavage cocktails all survive at low levels. Residual solvent testing appears on pharmaceutical-grade specifications and essentially never on retail research COAs. Its absence is not evidence of a problem; its presence is evidence of a more complete testing programme. For the manufacturing context these solvents come from, see our guide on how peptides are made.

Mass spectrometry: what it proves

Chromatography tells you how pure. Mass spectrometry tells you what it is — within limits that are important and rarely stated.

A mass spectrometer ionises the molecule, separates ions by mass-to-charge ratio (m/z), and counts them. Two ionisation methods dominate peptide work and they produce visibly different spectra.

Electrospray ionisation (ESI) sprays the sample from solution through a charged capillary, generating highly charged ion species by multiple protonation in positive mode — the number of charges tracking the number of basic sites available.^[12] The result is not one peak but a charge-state envelope: the same molecule appearing repeatedly at [M+2H]²⁺, [M+3H]³⁺, [M+4H]⁴⁺ and so on, each at a different m/z. Because those are all the same molecule, the mass can be reconstructed from any two adjacent peaks, and that redundancy is a genuine strength: several independent estimates of the same number.^[12] Software collapses the envelope into a single deconvoluted mass, which is the figure quoted on the COA.

Matrix-assisted laser desorption/ionisation (MALDI) co-crystallises the sample with a UV-absorbing matrix and fires a laser at it. Single protonation is the most frequent outcome, so the dominant signal is [M+H]⁺ and the mass is read almost directly.^[12] Spectra are simpler to look at and MALDI tolerates salt and buffer better, which is convenient for a lyophilised salt straight from the vial.

Then there is the monoisotopic-versus-average distinction, which trips up more COA readers than any other line. Monoisotopic mass is calculated using the exact masses of the most abundant isotope of each element — ¹H at 1.007825, ¹²C at exactly 12.000000. Average mass uses the isotope-weighted average atomic masses — hydrogen 1.00794, carbon 12.011. The two diverge more as the molecule gets larger and carbon-richer.^[12] For a small peptide, high-resolution instruments resolve individual isotope peaks and the monoisotopic value is quoted; for larger species the isotope envelope is unresolved and the average mass is what is reported. A COA that quotes an observed mass without saying which convention it used has left the reader unable to judge whether a small discrepancy is an error or a definition.

Mass accuracy expectations are well established. With modern instruments, peptides measured by either ESI or MALDI are typically accurate to 10–50 ppm; purified proteins by ESI under denaturing conditions to about 200 ppm; MALDI with delayed extraction above 25 kDa considerably worse, at 100–1000 ppm; and FT-ICR instruments to around 3 ppm with external calibration.^[12] A COA matching observed to theoretical mass to four decimal places on a nominal-mass instrument is claiming precision the instrument class does not deliver.

The two methods are complementary and neither substitutes for the other. That is not a slogan; it is a structural fact about what each instrument physically measures. Pharmacopoeial specification guidance makes the same point from the regulatory side: identification testing must be specific for the substance, and “identification solely by a single chromatographic retention time, for example, is not regarded as being specific” — acceptable approaches include infrared spectroscopy, two chromatographic procedures using different separation principles, or a combined technique such as HPLC/MS.^[6] A recent review of peptide quality assessment puts the same requirement more directly: identity should be established by at least two orthogonal techniques.^[1]

What actually proves sequence

A single intact-mass measurement is a weaker claim than it looks. It establishes that the molecule weighs what the claimed sequence should weigh. Several distinct failure modes survive that test intact:

• Scrambled sequence. The same residues assembled in the wrong order weigh exactly the same.
• Isobaric substitution. Leucine and isoleucine are structural isomers with identical mass. No mass measurement at any resolution can separate them.
• Compensating substitutions. Two changes whose mass differences cancel leave the total unchanged.
• Epimers. A D-residue where an L-residue belongs has identical elemental composition and identical mass.

Two techniques address this properly.

Tandem mass spectrometry (MS/MS). The intact peptide ion is isolated and then fragmented, most often along the amide backbone. The resulting fragment ions retaining the N-terminus are the b ions; those retaining the C-terminus are the y ions. Because each successive b ion differs from the last by exactly the residue mass of one amino acid, reading the spacings along the ladder reads the sequence. This is the basis of peptide mapping, the pharmacopoeial approach to characterising primary structure in biotechnology-derived articles.^[13] It resolves scrambling and compensating substitutions cleanly. It does not resolve leucine from isoleucine, and it does not resolve D from L, since fragment masses are identical in both cases.

Edman degradation. The chemical method that founded protein sequencing, in which the N-terminal residue is derivatised, cleaved and identified one cycle at a time. It has been the classical standard since Pehr Edman's 1949 description of the method,^[14] and it fails on peptides with a blocked or modified N-terminus — which includes any N-acetylated or pyroglutamate-capped species — while being slow and comparatively sample-hungry. Mass spectrometry displaced it for most purposes, but for unambiguous N-terminal confirmation it remains meaningful.

Amino acid analysis provides a third, weaker check: it confirms composition, not order. If the hydrolysate shows four arginines where the sequence calls for three, something is wrong even if the mass happened to work out.

The honest position: most retail peptide COAs include none of these. An intact mass and a chromatogram is the norm. That is not automatically a problem — for material from a supplier with a consistent process, intact mass plus a clean chromatogram plus a retention time consistent with previous lots is reasonable evidence. But it is worth knowing precisely which claim you are being sold. A COA that includes an MS/MS fragment table or a peptide map is making a materially stronger claim than one that does not, and it is fair to expect that on anything unusual, novel, or expensive.

The impurities that matter

The impurities in a synthetic peptide are not random contamination. They are a predictable family of near-relatives of the target, generated by known failure modes in synthesis, cleavage, purification and storage. Knowing the family tells you what those small peaks beside the main peak are likely to be.

Empirical evidence of the scale is useful. A 2025 study characterising a commercial peptide pool nominally containing 32 peptides, using UHPLC with high-resolution Orbitrap mass spectrometry, UV detection at 214 nm and a 5 ppm mass tolerance, identified 37 impurities — more impurity species than nominal components: 23 homo- and heterodimers from free cysteines, 6 methionine sulfoxides, 4 N-terminal pyroglutamate conversions, 2 deletion peptides, one aspartimide and one deamidation product. Residual parent peptide in the cysteine-containing peptides ranged from 71% to 94%, and one N-terminal glutamine peptide had converted 24% to pyroglutamate.^[15]

Three of these deserve elaboration because they are the ones that most often hide.

Deamidation converts an asparagine or glutamine side-chain amide to a carboxylic acid, adding 0.98 Da. That is a small enough shift that a low-resolution instrument may not separate it from the parent isotope envelope, and chromatographically it frequently appears as peak asymmetry rather than as a distinct peak. Rates are sequence-dependent, driven by the residue immediately following the asparagine, and are influenced by pH, temperature and local structure.^[17] It is also a shelf-life process, not only a synthesis one: a peptide can pass a COA and deamidate afterwards.

Aspartimide formation is the base-mediated cyclisation of an aspartate side chain onto the following backbone nitrogen, and it is a signature Fmoc-chemistry problem because repeated piperidine deprotection supplies exactly the base required. What makes it nasty analytically is that a single event generates a cluster of related products — the α-linked and β-linked ring-opened peptides plus their D-epimers — which elute close together and close to the parent.

Epimerisation is the most invisible of all. The dominant mechanism is oxazolone formation following strong activation of the carboxyl group, with direct α-proton abstraction by base as a secondary route.^[16] The product has identical molecular formula and identical mass, so mass spectrometry is blind to it in every mode including MS/MS. Only chromatography can resolve diastereomers, and epimerised products are so similar to the parent that separating them is difficult even deliberately.^[16] An epimer peak that co-elutes is counted as purity.

The general pattern behind the whole table: the impurities that are hardest to remove are the ones that most resemble the target, because chromatographic separation depends on physical difference. Purification review literature makes this explicit — species differing from the target by a single amino acid or a single functional group are the ones that resist separation by a single chromatographic mode.^[9] So the residual impurity profile is systematically enriched in exactly the species a single-mode HPLC purity assay is worst at detecting. That is not a scandal. It is a structural property of the measurement, and the reason orthogonal methods exist.

Endotoxin, bioburden and sterility

Chemical purity says nothing about biological contamination. These are separate tests and they appear on a minority of research-grade COAs.

Bacterial endotoxin is lipopolysaccharide from the outer membrane of Gram-negative bacteria. It survives autoclaving, it is not removed by sterile filtration, and it is biologically active at very low concentrations in many cell-based systems — which makes it a genuine confounder in in-vitro work regardless of any other consideration. The compendial method has historically been the Limulus amoebocyte lysate (LAL) assay, in gel-clot, turbidimetric and chromogenic formats, described in the bacterial endotoxins test chapter.^[25]

Recombinant alternatives have now been formalised. A separate general chapter covering the bacterial endotoxins test using recombinant reagents — recombinant Factor C (rFC) and recombinant cascade reagents (rCR) — addresses their use, validation and comparability.^[26] FDA's revised guidance on pyrogen and endotoxins testing removed certain LAL-specific references to accommodate a broader scope of recombinant reagents, while placing responsibility on the user to verify the assay is suitable for its intended purpose.^[27]

Bioburden counts viable organisms. Sterility is a different and stronger claim, requiring validated aseptic processing. Lyophilised research peptide is generally neither sterile nor claimed to be, and a COA asserting sterility without a named method and a named laboratory is asserting something expensive that is easy to type.

The 2024 online-market study cited earlier is instructive here too: endotoxin was detected in all three delivered vials at 2.16–8.95 EU/mg, even though no viable microorganisms were recovered at the time of testing.^[28] Absence of growth is not absence of endotoxin. They are separate questions with separate assays, and a COA that answers neither has not addressed the topic.

The standards a COA is measured against

Research-grade peptides are not sold as pharmaceuticals and are not required to meet pharmaceutical specifications. But the pharmaceutical framework is the only well-developed vocabulary for peptide quality, and knowing it tells you what a COA is implicitly comparing itself to.

There is a specific pharmacopoeial general chapter for this material class — Quality Attributes of Synthetic Peptide Drug Substances — alongside a companion chapter on starting materials for chemical peptide synthesis.^[2],[1] Around them sit the general chapters for the individual measurements: chromatography,^[11] water determination,^[23] acetic acid in peptides,^[22] trifluoroacetic acid in peptides,^[21] amino acid analysis,^[18] peptide mapping,^[13] and the endotoxin chapters.^[25],[26]

On impurity thresholds there is a subtlety worth spelling out, because it is frequently misquoted. The general ICH guideline on impurities in new drug substances sets tiered thresholds — reporting at 0.05%, identification at 0.10% and qualification at 0.15% in its principal tier, with lower figures where clinical exposure to the substance is higher.^[7] Those figures are widely repeated in peptide contexts — but the same guideline explicitly excludes peptides from its scope, alongside biological, oligonucleotide, radiopharmaceutical, fermentation and semi-synthetic products.^[7] Anyone citing ICH Q3A thresholds as the governing standard for a peptide is citing a document that says it does not apply.

The peptide-specific numbers come from elsewhere. FDA's 2021 guidance on abbreviated applications for certain highly purified synthetic peptide drug products recommends identifying peptide-related impurities at 0.10% of the drug substance or greater, indicates new peptide-related impurities generally should not exceed 0.5% for that pathway, requires impurities shared with the reference product to be present at levels the same as or lower than in the reference, and recommends high-resolution procedures such as UHPLC-HRMS.^[4] Agency training material frames it as: above 0.5% is not acceptable; between 0.10% and 0.5% must be identified, characterised and justified, with immunogenicity risk assessed.^[5] European practice applies report / identify / qualify levels of 0.1% / 0.5% / 1.0%,^[1] and the EU regulator adopted a dedicated synthetic-peptide guideline, EMA/CHMP/CVMP/QWP/367182/2025, on 9 December 2025, effective 1 June 2026.^[3]

One more framework belongs here because it bears on the laboratory rather than the molecule. ISO/IEC 17025:2017, third edition, published November 2017, sets general requirements for the competence of testing and calibration laboratories, covering competence, impartiality and consistent reporting of results, and it is the standard laboratory accreditation bodies assess against.^[31] When a COA names a laboratory, whether that laboratory holds accreditation — and whether the specific test is within its accredited scope, which is not the same thing — is checkable in public accreditation registries. That check takes about two minutes and is the highest-value verification step available to a non-specialist.

Method validation is the other half. The current ICH guideline on validation of analytical procedures, adopted 1 November 2023, sets out the characteristics an analytical procedure should demonstrate: specificity/selectivity, accuracy, precision at repeatability and intermediate levels, an appropriate range and response model, detection and quantitation limits (conventionally at signal-to-noise of about 3:1 and 10:1 respectively), and robustness. For chromatographic separations it specifies that suitable discrimination be shown at an appropriate level, for instance by resolution of the two closest-eluting components.^[8] A purity method that has never been shown to resolve the closest-eluting impurity is a method whose purity number is undefined at the level that matters.

A worked example, line by line

What follows is an illustrative certificate constructed for teaching. The peptide is a generic 30-residue sequence, the lot code is invented, and the values were chosen to demonstrate specific reading points. It is not a Patriot Labs lot and it is not a record of any real test. Read it as a worked exercise.

The reading of that certificate in one sentence: the material is what it claims to be, it is chromatographically clean at a wavelength that would show most impurities, and roughly a sixth of the powder is salt and water — which is normal and is now known rather than assumed.

Now consider the same certificate with three lines deleted — no column, no wavelength, no content. Purity 98.4% and mass 3422.1 remain. Nothing on it is false. It has become almost uninterpretable. That deletion is the difference between most retail COAs and a useful one, and no fabrication was required to produce it.

How to read a COA, in order

A repeatable sequence. Each step is cheap; the early ones eliminate most bad documents before you spend attention on chemistry.

1. Match the lot. Read the batch code on the certificate and the batch code on the vial. If they do not match exactly, or the vial has no code, everything downstream is unanchored. This is the most common single failure and it takes ten seconds.
2. Find the date. A COA with no date of analysis is not a laboratory record. Note also how old it is; a certificate from two years ago describes material two years ago.
3. Identify the laboratory. Look for a name, an address, and if claimed, an accreditation number. Search it. A laboratory that does not exist online, or exists only as a logo, is a finding.
4. Check identity before purity. Confirm the observed mass against the theoretical mass for the stated sequence, and check that the certificate says whether the mass is monoisotopic or average and which ionisation method produced it. Wrong compound at high purity is still the wrong compound.
5. Read the purity method, then the purity number. Column, gradient, run time, wavelength. Without them the percentage is a number without units. With them you can judge whether it is a conservative measurement or a flattering one.
6. Look for the chromatogram and the spectrum. Actual traces, with legible axes, retention times consistent with the stated gradient, and a visible baseline. A number without a trace is an assertion.
7. Hunt for content, water and counterion. If present, do the mass balance. If absent, register that the purity figure is not a mass fraction and adjust your mental model of what is in the vial.
8. Check what is not tested. No endotoxin line means endotoxin is unknown, not absent. No sequence confirmation means sequence is inferred from mass. Note the silences.
9. Ask for anything missing. A supplier that can produce the full report for your lot on request is behaving like an operation that has one. A supplier that cannot, or that sends the same PDF regardless of lot, has answered a different and more important question.

How to spot a weak or fabricated COA

Two categories here, and it is worth keeping them apart. A weak COA is a real document that omits things. A fabricated COA is a graphic. The tells differ.

Three checks add disproportionate value. Try to select the text — instrument software exports real text layers, and a COA whose every character is baked into a bitmap has been through an image editor at some point (a scanned signature page is an innocent explanation, but notice it). Read the PDF properties — creation date, modification date, author and producing application; a report authored in a design application, or created months before the stated analysis date, is telling you something. Compare two lots of the same product — real batches differ, purity moves by tenths of a percent, retention times drift, minor peaks come and go; two lots with identical numbers to the decimal and a pixel-identical trace were not measured twice.

The market context makes this worth the effort. In the online-purchase study cited earlier, the delivered products carried vendor purity claims of 99% while measuring 7.7–14.37% by LC-MS.^[28] Claims are cheap. Our guide to buying research peptides covers the broader supplier-evaluation problem this sits inside, and the supplier comparison guide looks at how different vendors handle documentation.

Third-party versus in-house testing

A seller grading its own product has an obvious conflict of interest. Independent testing removes it. That much is uncontroversial, and it is why third-party testing is worth asking for.

But the phrase “third-party tested” is not self-verifying, and it is worth being precise about what independence actually buys.

What it adds: a party with no financial stake in the result generated the number. If that party is accredited, it operates under documented requirements for competence and impartiality and reports results consistently.^[31] The report is checkable against a laboratory that exists and can be contacted.

What it does not add: independence does not fix a bad method. A third-party laboratory running a twelve-minute gradient at 280 nm reports a flattering number just as an in-house one would. It does not guarantee the sample was drawn representatively from the batch you received — a laboratory tests what it is sent. And it does not by itself mean the specific test falls within the laboratory's accredited scope, which is a separate line item worth reading.

The regulated-industry answer to this problem is instructive. Under US drug GMP rules, a manufacturer may accept a supplier's report of analysis in place of its own full testing — but only if it performs at least one specific identity test itself, and validates the supplier's test results at appropriate intervals.^[30] In other words: even inside a regulated supply chain with contracts and audits, a supplier's certificate is not accepted as sufficient on its own. It is accepted plus independent identity confirmation plus periodic verification that the supplier's numbers hold up.

That is a good model for a reader with no regulatory leverage. Treat a COA as strong evidence, not as proof. Prefer suppliers who name their laboratory. Prefer certificates that carry full method parameters and actual traces. And treat consistency across lots over time as its own form of evidence — a supplier producing per-lot reports with plausible batch-to-batch variation is demonstrating a process, which is more informative than any single document.

At Patriot Labs, batches are sent to an independent US laboratory for HPLC and mass-spectrometric analysis and the report is tied to the batch code on the vial. We would rather you check that than take our word for it, which is the entire point of the section you just read. Our about page covers how we source and test.

What a COA cannot tell you

Every section above describes what a certificate proves. This one describes what it does not, and it is the section that most COA guides skip because it is unflattering to the document being explained.

A COA is a snapshot, not a warranty. It records results from one sample, drawn from one batch, at one moment — usually shortly after manufacture. It does not describe the vial in your hand. It describes a sibling of that vial, at a point in the past.

It says nothing about what happened afterwards. Between the test date and the vial arriving, the material was filled, capped, stored, packed and shipped. Oxidation of methionine and tryptophan, deamidation of asparagine and glutamine, and hydrolysis all proceed in the solid state, and all accelerate with heat and moisture.^[17] A certificate cannot see any of it, which is why cold-chain handling and correct storage are what preserve the state the COA documented.

Sampling is an assumption. The certificate reports on the aliquot that reached the instrument. Whether that aliquot represents the batch depends on how it was drawn, and no COA documents that.

Purity is method-relative. As established above, the same vial yields different purity figures under different wavelengths, gradients and run times. There is no single true purity number, only a number produced by a stated method — which is why the method matters as much as the result.

Identity by mass is not identity by sequence. Unless MS/MS or Edman data is present, the sequence is inferred, not demonstrated.

Absence of a test is not a passing result. No endotoxin line means endotoxin was not measured. No residual solvent line means residual solvent was not measured. A short certificate is not a clean one.

And a COA says nothing about biology. It is a chemistry document. Purity, identity and content describe what a molecule is. They make no statement about what it does in any system, and nothing on a certificate should be read as one.

Once reconstituted, the clock changes again: material in solution is far less stable than lyophilised powder, and the diluent matters — see bacteriostatic water for that side of the problem. A COA is the starting condition. Everything after it is handling.

Frequently asked questions

What is a peptide COA?

A peptide COA, or Certificate of Analysis, is a batch-specific laboratory report stating what a single lot of peptide was tested for, which analytical methods were used, and what results those methods returned. A useful COA answers two questions: is the material the compound it claims to be (identity, usually by mass spectrometry), and how much of the material is that compound (chromatographic purity by RP-HPLC, and separately net peptide content). It is valid only for the lot number printed on it.

What does COA mean for peptides?

COA stands for Certificate of Analysis. For peptides it means a document tying a specific batch to specific test results — typically an RP-HPLC chromatogram with a purity percentage, a mass spectrum with an observed versus theoretical mass, and often water content, counterion content and appearance. The word certificate is doing less work than it sounds like: a COA certifies only that the named laboratory ran the named tests on the named lot and got the printed numbers.

How do you read a certificate of analysis for peptides?

Read it in order. First check the lot number matches the vial and that a test date is present. Then confirm identity: observed mass against theoretical, with ionisation mode and charge state stated. Then read purity: the RP-HPLC figure, the detection wavelength, the gradient and run time, and whether a chromatogram is attached. Then look for net peptide content, water and counterion, which say how much of the powder is peptide at all. Finally check who ran the tests.

What does 99% purity mean on a peptide COA?

Almost always it means chromatographic area purity: the main peak accounted for 99% of the total integrated UV peak area in that one RP-HPLC run, at one detection wavelength. It is a relative measure of peptide-related species, not a statement about mass. It excludes salt, water and residual solvent, assumes every species absorbs UV equally, and cannot see anything co-eluting with the main peak. A 99% purity figure is compatible with the powder being far less than 99% peptide by weight.

Is HPLC purity the same as peptide content?

No, and this is the single most important thing a COA usually buries. HPLC purity is the proportion of detected peptide-related material that is the target peptide. Net peptide content is the proportion of the physical powder that is peptide at all, with counterion salt, bound water and residual solvent subtracted out. Because peptides purified by preparative RP-HPLC are isolated as trifluoroacetate salts — a burden measured as high as 35% of total weight in one published study^[19] — and lyophilised solids hold water, content is routinely much lower than purity. Both numbers can be honest at once.

What does the mass spectrometry section of a peptide COA show?

It shows the measured molecular mass of the main species against the mass calculated from the claimed sequence. Electrospray ionisation produces a series of multiply charged ions that is deconvoluted back to a single mass; MALDI usually produces a singly charged ion read directly.^[12] A credible entry states the ionisation method, whether the mass is monoisotopic or average, and the observed value. Mass confirms molecular weight only — it cannot distinguish a scrambled sequence or an isobaric substitution from the correct peptide.

Does a COA prove the peptide sequence is correct?

Not on its own. A single intact mass measurement proves the molecule weighs what the claimed sequence should weigh. Any rearrangement of the same residues, any pair of substitutions that cancel out in mass, and any D-amino-acid epimer all share the same molecular mass. Sequence is proved by tandem mass spectrometry, where the molecule is fragmented and the b and y ion ladder is read, or by Edman degradation. Most retail peptide COAs include neither.

How can you tell if a peptide COA is fake?

Look for missing traceability and missing method detail. Warning signs include no lot number or one that does not match the vial, no test date, no named laboratory or a laboratory with no verifiable existence, a purity figure with no chromatogram attached, an identity claim with no spectrum, no column, gradient or wavelength stated, the same chromatogram image reused across different products, axis labels or retention times that do not match the stated method, and suspiciously round numbers such as 99.9% on every product in a catalogue. Checking the PDF's own metadata and comparing two lots of the same product are both fast and revealing.

Should a peptide COA come from a third-party lab?

Independent testing removes the obvious conflict of interest in a seller grading its own product, and an accredited laboratory operating to ISO/IEC 17025 has documented requirements for competence and impartiality.^[31] But third-party is not magic. The value comes from the laboratory being named and verifiable, the report carrying full method parameters, and the lot on the report matching the lot in hand. A third-party report with none of those attributes is worth no more than an in-house one.

Does a COA guarantee what is in the vial I received?

No. A COA is a snapshot of one batch at one moment, usually taken shortly after manufacture, on a sample drawn from that batch — not on your vial. Nothing on the certificate accounts for what happened during filling, shipping and storage afterwards. Peptides oxidise, deamidate and hydrolyse over time, and heat and moisture accelerate all three. A COA tells you the material was good when tested; storage history determines whether it still is.

All Patriot Labs products are sold strictly for in-vitro research and laboratory use only. Not for human or veterinary consumption. This guide is educational and describes analytical chemistry and published standards in general terms; it is not medical advice, does not describe how to use any product, and the references cited do not constitute a product claim.

For in-vitro research and laboratory use only. Not for human consumption.