How to read a certificate of analysisWhat the studies show

Q: What kind of evidence exists for certificate of analysis?

Mixed evidence. Reference standards and pharmacopoeial methods rather than studies; the interpretation rests on what each test does and does not measure.

A certificate of analysis is a lab's record of what it measured in one sample on one day. Every line answers a specific question, and a few are often missing.

5 min read · Mixed evidence · Updated November 2026

In short

Name: certificate of analysis
Evidence: Mixed evidence
Bottom line: Reference standards and pharmacopoeial methods rather than studies; the interpretation rests on what each test does and does not measure.

Before you read this

This guide explains what published studies found. That is all it is.

It is not medical advice and it is not instructions for using anything. Where a dose appears, we are reporting what a named study gave the people or animals in it — a fact about the study, not an instruction to you.

The compounds here are not approved by the FDA for use in people and are sold for laboratory research only. A study tells you what happened to the subjects in it; it does not tell you what would happen to you. Talk to a doctor first.

What it is

A certificate of analysis is a document a testing laboratory issues to say what it measured in a sample, how, and with what result. The form is defined in manufacturing guidance: the ICH Q7 guideline, which governs the making of active pharmaceutical ingredients, says a certificate should list each test performed, the acceptance limit, the numerical result, the batch it came from and the date, and be signed by the laboratory ¹. That definition is the standard to read any certificate against.A certificate that lacks a method, a date, a reference you can tie back to the sample (a batch or sample number, or a verification key the laboratory will confirm), or the laboratory's own name is missing the parts that make it checkable.

The rest of this guide is free

Unlock the How to read a certificate of analysis guide — and every other one

What the studies found, what they didn't, every source linked. Enter your email once; the link we send opens all of them on this device, plus one plain-language study breakdown a week.

The lines, one by one

Identity. The first question is whether the material is the molecule named on the label. The usual answer is a mass spectrum: the instrument weighs the molecule, and the certificate reports the observed mass next to the mass calculated from the sequence, which should agree to within a fraction of a unit ². A chromatogram alone does not establish identity, because it only shows that most of the material behaves alike on a column.

HPLC purity. The share of the chromatogram's total peak area that belongs to the main peak, as a percentage. It measures how much ultraviolet-absorbing impurity is present under that laboratory's conditions, and a serious certificate names the column, gradient and wavelength it used, as the pharmacopoeial chromatography chapter USP <621> requires for a reportable result ³. It is not a weight percentage: salts and water do not appear on it.

Appearance. A visual description — "white lyophilised powder" — recorded because a change in colour or form between lots is a cheap early warning. It is the least informative line on the page and the easiest to pass.

Endotoxin. Reported in endotoxin units per millilitre or per vial, from the Limulus amebocyte lysate assay or its recombinant equivalent under USP <85>. Endotoxin is the fever-causing residue of bacterial cell walls; it survives filtration, and it is the main reason a chemically pure peptide can still fail a release specification ³.

Sterility. A pass or fail from USP <71>, in which the sample is incubated in two culture media for 14 days and watched for growth ³. Sterility and endotoxin are different tests for different things: a sample can be sterile and still carry endotoxin from bacteria that were killed earlier.

Water, net content, and what is often left out

Two lines appear on a pharmaceutical certificate and are often absent from a research one. Water content, by Karl Fischer titration under USP <921>, says how much moisture the solid holds, which matters because moisture drives degradation ³. Net peptide content, by amino acid analysis or nitrogen determination, says what fraction of the powder's weight is peptide as opposed to counter-ion and water; it is the number that connects a stated milligram figure to the amount of molecule actually present ². When those lines are missing, the certificate is answering the question of purity and leaving the question of quantity open.

Who tested it, and when

The laboratory's name, its accreditation and the test date carry as much weight as the numbers. Accreditation to ISO/IEC 17025 means an outside body has audited the lab's methods, equipment and records, and it is the standard that regulators recognise for a testing laboratory ⁴. The date ties the result to a moment: a purity figure is a measurement of the sample as it was that day, and a peptide stored badly afterwards is not the sample that was measured. A verification key — a certificate number, a QR code, or a lot-specific page on the laboratory's own site — exists so that the document can be checked against the lab's record rather than taken on the strength of a PDF, since a certificate is a file and files are easy to edit.

What a certificate is not

It is not a guarantee. It reports results on a sample drawn from a lot; the assumption that the rest of the lot matches rests on the maker's process being consistent, which the certificate does not test. It is not a statement about any vial after the test date. It makes no claim about what the material is suitable for; it says what was measured. A 2014 review of impurities in peptide medicines is a useful companion to any certificate, because it lists the impurities that routine HPLC may not separate and makes clear that a high purity figure is a strong statement about one class of contaminant and a silent one about the others ².

What these studies are, and aren't

Mixed evidence. Reference standards and pharmacopoeial methods rather than studies; the interpretation rests on what each test does and does not measure.