A peptide certificate of analysis looks official because it is full of percentages. That is also how it fools people. A COA is a lab report on one lot. It answers a short list of analytical questions. It does not approve a drug, it does not certify a protocol, and it does not prove that the vial in your rack is the lot on the PDF unless the identifiers match.
If you can separate the questions the methods actually answer, you can read almost any peptide COA without trusting the header art. The rest of this page is that separation: identity versus purity, what HPLC cannot see, what mass spectrometry cannot quantify, and why endotoxin is a different test from both.
First check: is this paper about this vial?
Before HPLC, look at the boring fields. If these do not match, the rest of the document is about someone else’s box.
Lot or batch number on the vial versus the COA
Peptide name and, where given, CAS or sequence
Labeled amount (for example 10 mg) versus any net-content result
Test date and the name of the laboratory that ran the work
“Batch-matched” means the certificate was issued for the lot printed on that vial, not for a cousin lot from the same catalog page. A generic “typical purity ≥99%” banner is not a COA.
Also notice who signed it. An in-house printout from the same organization that filled the vial is a different kind of document from a report on another laboratory’s letterhead, with a method, a chromatogram, and an accession or sample ID. Third-party does not mean infallible. It means the tester and the seller are not the same inbox. If the lab name is missing, you cannot even ask that question.
The two questions people collapse into one number
Identity: is the main species the molecule on the label?
Purity: of the material the method can see, how much of the peak area is that main species?
A sample can fail one and pass the other. A 99% HPLC chromatogram of the wrong peptide is a clean wrong answer. A correctly identified peptide at 92% HPLC is a different kind of problem. You need both, preferably from orthogonal methods, which is the unglamorous word for “not the same trick twice.”
Test
Question it answers
Question it does not answer
HPLC (often RP-HPLC)
Relative peak-area purity at the detection wavelength
Whether the main peak is the correct sequence; endotoxin; water; many salts
Mass spectrometry
Whether the observed mass matches the theoretical mass of the claimed peptide
A purity percentage comparable to HPLC area-%
Net peptide content (when present)
How much of the weighed solid is peptide rather than counter-ion, residual solvent, or water
Sequence identity by itself
Bacterial endotoxins (LAL or rFC)
Lipopolysaccharide contamination from Gram-negative bacteria, as the assay detects it
Identity or HPLC purity
HPLC purity: a percentage of what the detector saw
High-performance liquid chromatography separates components as they move through a column. For peptides, reverse-phase HPLC (RP-HPLC) on an alkyl-bonded phase (often C18) is the usual workhorse. United States Pharmacopeia general chapter <621> is the chromatography reference laboratories point to for system setup and calculations.
The purity number on most peptide COAs is area percent: the area of the main peak divided by the total integrated peak area, at a stated wavelength. It is not “99% of the milligrams in the bottle.” It is “99% of the UV-visible, chromatographically resolved, peptide-related peak area under this method.”
That definition has consequences:
Wavelength matters. Aromatic residues absorb strongly around 280 nm. Peptides with no aromatic residues are often watched near 210-220 nm, where peptide bonds absorb. A method that is blind to a deletion impurity at the chosen wavelength will not subtract it.
Co-elution inflates purity. Two species that do not separate under this gradient are counted as one peak. Purity looks better than the chemistry.
Non-peptide mass is often invisible. Water, acetate or TFA counter-ions, and many residual solvents do not show up as peptide peaks. A vial can be “99% by HPLC” and still be a smaller fraction peptide by weight.
The main peak is unnamed by HPLC alone. Retention time is not a sequence. You still need an identity method.
A serious COA prints a chromatogram, or at least the method: column, gradient, wavelength, and which peak was integrated as the analyte. A number with no method is a number you cannot check.
When a chromatogram is included, read it like a picture, not like a logo. You want a dominant peak, a baseline that is not a forest of junk, and integration marks that match the reported %. A cropped snippet that shows only the main peak, with no time axis and no minor peaks, is not a chromatogram you can interrogate. If the method lists a gradient and the PDF has no trace, you are being asked to trust a calculator without the input.
Related-substance peaks (deletion sequences, truncated chains, oxidation products) often sit near the main peak. That is why the method has to separate them. A short isocratic run that parks everything in one lump will print a flattering area-%.
Mass spectrometry: identity by mass, not a second purity score
Mass spectrometry ionizes the peptide and measures mass-to-charge. Electrospray (ESI) and MALDI-TOF are the two you will see most. USP general chapter <736> covers mass spectrometry as a pharmacopeial technique. On a peptide COA you want a theoretical mass (from the sequence, including the correct disulfide, amidation, acetylation, or lipid if claimed) and an observed mass.
Agreement within the instrument’s stated accuracy is evidence that the main species has the right empirical formula in that mass window. It catches many synthesis failures: missing residue, extra residue, wrong protecting-group leftover that changes mass. It does not, by itself, tell you the area-% of related impurities. A wrong peptide with the same nominal mass (rare, but isomers and isobaric swaps exist) is why some labs add MS/MS or amino-acid analysis when the lot is a standard rather than a catalog reagent.
If the COA has HPLC and no mass, you have a clean chromatogram of an unnamed peak. If it has mass and no HPLC, you have a mass that might belong to a minority species. Together they are the minimum pair that makes the word “purity” mean something next to the word “identity.”
Net content vs the number on the cap
Lyophilized peptides are often isolated as salts. Trifluoroacetate and acetate are common counter-ions after HPLC purification. The cake also holds residual water. HPLC area-% does not subtract those. Net peptide content (sometimes from nitrogen analysis or amino-acid analysis) asks what fraction of the weighed solid is peptide.
That is why a 10 mg labeled vial and a 10 mg “as-is” weigh-out are not automatically 10 mg of peptide analyte for a quantitative assay. If the COA reports net content, use it for the math in the lab notebook. If it does not, you do not get to invent a percentage. You only know you lack the number.
Trifluoroacetate (TFA) is the usual leftover from reverse-phase purification in TFA-containing mobile phases. Acetate lots exist when the lab exchanged the counter-ion. Neither counter-ion is “the peptide.” Both add mass to the cake. If the COA names the salt form, you can at least know which invisible partner is in the vial. If it does not, HPLC % still does not include it.
Blends make this sharper. A GLOW or KLOW-style mixture has more than one analyte. A single HPLC % for the whole cake is weaker than per-component identity plus per-component milligrams. Nickname blends are a good stress test of whether the COA is actually about composition. See GLOW vs KLOW composition and naming.
Endotoxin is not a peak on the peptide chromatogram
Bacterial endotoxins are lipopolysaccharides from Gram-negative outer membranes. They are not peptides. They do not appear as a peptide-related HPLC peak, and they do not report as the peptide’s molecular ion. A lot can be high by HPLC, correct by mass, and still carry endotoxin if that test was never run.
The pharmacopeial bacterial endotoxins test is USP <85>, classically the Limulus amebocyte lysate (LAL) assay. A recombinant Factor C method is described in USP <86>. Results are typically in endotoxin units (EU) per mass or per vial.
A pass/fail line copied from an injectable drug monograph does not automatically apply to a research reagent, and this page will not invent one. What you can say as a reader: if endotoxin matters for the in-vitro system (many cell assays), look for a reported value and a method, not a vibe.
Other lines you may or may not get
Research-peptide COAs are not ICH Q6A drug-substance specifications. Extra fields are useful when present, and their absence is information too:
Appearance of the lyophilized solid
Solubility in a stated solvent (a lab observation, not a use instruction)
Water (Karl Fischer) when residual moisture was measured
Residual solvents
Counter-ion identity
Amino-acid analysis or peptide mapping on higher-grade lots
Sterility (USP <71>) is a different claim from endotoxin. A non-sterile powder can still have a low endotoxin result, and a “sterile” claim without a method is just a word.
Copper-peptide lots (GHK-Cu, and blends that contain it) sometimes list color. The copper(II) complex is a distinct analyte from metal-free GHK. If the COA identity is GHK-Cu, the mass and the HPLC method should be the complex, not the apo-peptide. A blend COA should not hide that as one anonymous “purity” line. Nickname catalogs make this failure mode common, which is why composition names and COA lines have to be read together.
When fields disagree, do not average them. A mass that matches peptide A and an HPLC label that says peptide B is a failed identity, not a 50/50. A lot number that does not match the vial is not a COA for that vial, even if the peptide name is right. Those are stop conditions, not puzzles.
A short reading order
Match lot numbers.
Read the identity method (mass, and sequence or CAS if given).
Read HPLC with the method attached, not the percentage in isolation.
Look for net content if you need a mass of peptide, not a mass of cake.
How to Read a Peptide Certificate of Analysis (COA)
A peptide certificate of analysis looks official because it is full of percentages. That is also how it fools people. A COA is a lab report on one lot. It answers a short list of analytical questions. It does not approve a drug, it does not certify a protocol, and it does not prove that the vial in your rack is the lot on the PDF unless the identifiers match.
If you can separate the questions the methods actually answer, you can read almost any peptide COA without trusting the header art. The rest of this page is that separation: identity versus purity, what HPLC cannot see, what mass spectrometry cannot quantify, and why endotoxin is a different test from both.
First check: is this paper about this vial?
Before HPLC, look at the boring fields. If these do not match, the rest of the document is about someone else’s box.
“Batch-matched” means the certificate was issued for the lot printed on that vial, not for a cousin lot from the same catalog page. A generic “typical purity ≥99%” banner is not a COA.
Also notice who signed it. An in-house printout from the same organization that filled the vial is a different kind of document from a report on another laboratory’s letterhead, with a method, a chromatogram, and an accession or sample ID. Third-party does not mean infallible. It means the tester and the seller are not the same inbox. If the lab name is missing, you cannot even ask that question.
The two questions people collapse into one number
Identity: is the main species the molecule on the label?
Purity: of the material the method can see, how much of the peak area is that main species?
A sample can fail one and pass the other. A 99% HPLC chromatogram of the wrong peptide is a clean wrong answer. A correctly identified peptide at 92% HPLC is a different kind of problem. You need both, preferably from orthogonal methods, which is the unglamorous word for “not the same trick twice.”
HPLC purity: a percentage of what the detector saw
High-performance liquid chromatography separates components as they move through a column. For peptides, reverse-phase HPLC (RP-HPLC) on an alkyl-bonded phase (often C18) is the usual workhorse. United States Pharmacopeia general chapter <621> is the chromatography reference laboratories point to for system setup and calculations.
The purity number on most peptide COAs is area percent: the area of the main peak divided by the total integrated peak area, at a stated wavelength. It is not “99% of the milligrams in the bottle.” It is “99% of the UV-visible, chromatographically resolved, peptide-related peak area under this method.”
That definition has consequences:
A serious COA prints a chromatogram, or at least the method: column, gradient, wavelength, and which peak was integrated as the analyte. A number with no method is a number you cannot check.
When a chromatogram is included, read it like a picture, not like a logo. You want a dominant peak, a baseline that is not a forest of junk, and integration marks that match the reported %. A cropped snippet that shows only the main peak, with no time axis and no minor peaks, is not a chromatogram you can interrogate. If the method lists a gradient and the PDF has no trace, you are being asked to trust a calculator without the input.
Related-substance peaks (deletion sequences, truncated chains, oxidation products) often sit near the main peak. That is why the method has to separate them. A short isocratic run that parks everything in one lump will print a flattering area-%.
Mass spectrometry: identity by mass, not a second purity score
Mass spectrometry ionizes the peptide and measures mass-to-charge. Electrospray (ESI) and MALDI-TOF are the two you will see most. USP general chapter <736> covers mass spectrometry as a pharmacopeial technique. On a peptide COA you want a theoretical mass (from the sequence, including the correct disulfide, amidation, acetylation, or lipid if claimed) and an observed mass.
Agreement within the instrument’s stated accuracy is evidence that the main species has the right empirical formula in that mass window. It catches many synthesis failures: missing residue, extra residue, wrong protecting-group leftover that changes mass. It does not, by itself, tell you the area-% of related impurities. A wrong peptide with the same nominal mass (rare, but isomers and isobaric swaps exist) is why some labs add MS/MS or amino-acid analysis when the lot is a standard rather than a catalog reagent.
If the COA has HPLC and no mass, you have a clean chromatogram of an unnamed peak. If it has mass and no HPLC, you have a mass that might belong to a minority species. Together they are the minimum pair that makes the word “purity” mean something next to the word “identity.”
Net content vs the number on the cap
Lyophilized peptides are often isolated as salts. Trifluoroacetate and acetate are common counter-ions after HPLC purification. The cake also holds residual water. HPLC area-% does not subtract those. Net peptide content (sometimes from nitrogen analysis or amino-acid analysis) asks what fraction of the weighed solid is peptide.
That is why a 10 mg labeled vial and a 10 mg “as-is” weigh-out are not automatically 10 mg of peptide analyte for a quantitative assay. If the COA reports net content, use it for the math in the lab notebook. If it does not, you do not get to invent a percentage. You only know you lack the number.
Trifluoroacetate (TFA) is the usual leftover from reverse-phase purification in TFA-containing mobile phases. Acetate lots exist when the lab exchanged the counter-ion. Neither counter-ion is “the peptide.” Both add mass to the cake. If the COA names the salt form, you can at least know which invisible partner is in the vial. If it does not, HPLC % still does not include it.
Blends make this sharper. A GLOW or KLOW-style mixture has more than one analyte. A single HPLC % for the whole cake is weaker than per-component identity plus per-component milligrams. Nickname blends are a good stress test of whether the COA is actually about composition. See GLOW vs KLOW composition and naming.
Endotoxin is not a peak on the peptide chromatogram
Bacterial endotoxins are lipopolysaccharides from Gram-negative outer membranes. They are not peptides. They do not appear as a peptide-related HPLC peak, and they do not report as the peptide’s molecular ion. A lot can be high by HPLC, correct by mass, and still carry endotoxin if that test was never run.
The pharmacopeial bacterial endotoxins test is USP <85>, classically the Limulus amebocyte lysate (LAL) assay. A recombinant Factor C method is described in USP <86>. Results are typically in endotoxin units (EU) per mass or per vial.
A pass/fail line copied from an injectable drug monograph does not automatically apply to a research reagent, and this page will not invent one. What you can say as a reader: if endotoxin matters for the in-vitro system (many cell assays), look for a reported value and a method, not a vibe.
Other lines you may or may not get
Research-peptide COAs are not ICH Q6A drug-substance specifications. Extra fields are useful when present, and their absence is information too:
Sterility (USP <71>) is a different claim from endotoxin. A non-sterile powder can still have a low endotoxin result, and a “sterile” claim without a method is just a word.
Copper-peptide lots (GHK-Cu, and blends that contain it) sometimes list color. The copper(II) complex is a distinct analyte from metal-free GHK. If the COA identity is GHK-Cu, the mass and the HPLC method should be the complex, not the apo-peptide. A blend COA should not hide that as one anonymous “purity” line. Nickname catalogs make this failure mode common, which is why composition names and COA lines have to be read together.
When fields disagree, do not average them. A mass that matches peptide A and an HPLC label that says peptide B is a failed identity, not a 50/50. A lot number that does not match the vial is not a COA for that vial, even if the peptide name is right. Those are stop conditions, not puzzles.
A short reading order
If you want lot PDFs collected in one index rather than a method explainer, that lives at the certificates of analysis index.
References
For laboratory research only. Not for human or veterinary use. A COA is an analytical report on a lot, not authorization for use in humans or animals.
Last reviewed: 2026-10-14