Peptide test accuracy: what ± and k = 2 on a COA tell you, and how repeatable a result is

How accurate a peptide test can be

No lab measures the exact amount of peptide in a vial. Every result is an estimate, and its accuracy is expressed as a measurement uncertainty: a range around the result that should contain the true value. The Guide to the Expression of Uncertainty in Measurement (GUM, JCGM 100:2008) treats a result as complete only when it carries that statement.

The range depends on the test. HPLC purity is a ratio of peak areas inside one run, so it needs no standard of known content, though integration settings still move it. Content in mg per vial is an absolute amount, measured against a reference standard, and it often moves more. In May 2026, Peptide Critic compared reports from five labs on one 10 mg retatrutide kit. Every purity result fell between 99.74% and 99.93%. The mg results ran from 10.18 to 12.80 mg, a spread of about 26%.

ICH Q2(R2) and the GUM separate three terms that reports often mix up:

Term	Plain meaning
Accuracy	How close a result is to the true or accepted reference value
Precision	How much repeated results scatter
Measurement uncertainty	The range around one result that reflects its known sources of error

What ± and k = 2 mean on a peptide COA

A lab first estimates the standard uncertainty, u, which is one standard deviation for the result. It then multiplies u by a coverage factor, k, to get the expanded uncertainty, U. The GUM states that with k = 2 the interval has a level of confidence of about 95%, when the distribution is close to normal. The Eurachem/CITAC guide for analytical labs recommends printing the result as x ± U, with the coverage factor named. It advises against ± for a standard uncertainty, because readers take ± to mean a high-confidence range.

An illustration, not a test result: “Content: 4.8 ± 0.3 mg per vial (k = 2)”. The lab expects the true content to lie between 4.5 and 5.1 mg. A 5 mg label claim sits inside that range, so the result does not show an underfill.

On a 10 mg result, margins that labs publish work out like this (our arithmetic):

Stated margin	On a 10 mg result	Range
±2%	±0.2 mg	9.8 to 10.2 mg
±5%	±0.5 mg	9.5 to 10.5 mg
±10%	±1.0 mg	9.0 to 11.0 mg

Repeatability, intermediate precision and reproducibility

ICH Q2(R2), the guideline for validating analytical methods for medicines, splits precision into three levels:

• Repeatability: the same conditions over a short time, checked with at least 9 determinations across the range, or 6 at the test concentration.
• Intermediate precision: one lab, across days, analysts and equipment.
• Reproducibility: between labs, from an inter-laboratory trial.

WHO’s good practices for quality control labs say that when investigative testing reports a concentration, the estimated uncertainty should be given too, from an uncertainty budget, validation data or proficiency tests. A vial bought on the open market is that kind of sample.

What a replicate repeats also matters. A second injection repeats only the instrument step. A second preparation also repeats the weighing and dilution. A second vial adds the fill. Five labs described their practice in answers to Finnrick’s method questionnaire:

Lab	Replicates and checks	Stated margin for mg	Rounding
Krause Analytical	One preparation, one injection, one vial by default; more on request	Check-standard acceptance of ±10%, which it treats as the lower limit of its error, including preparation and calibration	Two significant figures
BTLabs	No routine replicate injections; a weekly instrument check	±2–3%	1–2 significant figures
Chromate	Replicate injections of the standard only (RSD below 0.3%)	2–5%	Quantity to 4 significant figures; purity as the software reports it
Ichor Life Sciences	One injection; re-prepared or re-run if more than about 20% off the reference	No formal uncertainty established, as it states	2 decimal places
PharmLabs	Recovery evaluated during validation, not assumed	Expanded uncertainty at 95% confidence on each COA	Not checked

Reference standards set the scale for every mg result

Content per vial is measured against a reference standard of known content, so the standard’s assigned content sets the scale. The GUM lists inexact values of standards and reference materials as a source of uncertainty. If the content assigned to a standard is 5% too high, every result calculated from it reads about 5% high. Finnrick’s May 2026 blog describes one unnamed lab whose five reports were all off by the same proportion, a pattern that pointed to a calibration error.

Assigning that content is itself a measurement. USP scientists assign peptide reference standards by mass balance: they subtract HPLC impurities, acetate, TFA, residual solvents and residue on ignition. They measure water by Karl Fischer with vials opened under dry nitrogen, because peptides take up water from the air.

The basis matters as much as the standard. The first source of uncertainty the GUM lists is an incomplete definition of what is being measured. “10 mg” can mean free base, the salt or the whole powder. Our guide to net peptide content explains the gap.

Significant figures: how many digits a result deserves

The GUM says an uncertainty should usually be given to at most two significant digits, with the result rounded to match. So a result of 10.457 mg with an uncertainty of 0.4 mg should be printed as 10.5 ± 0.4 mg. The extra digits add no information.

Krause Analytical’s questionnaire notes that purity is often printed to four significant figures, which no commercial instrument can support, and estimates its own chromatographic values as valid to two. In our September 2026 review of public COAs from 25 labs and one rating site’s certificate wrapper, three decimals with no uncertainty were a common pattern. Of the 26, 21 printed mg per vial, but only 1 printed an expanded uncertainty with its coverage factor. Our report on 99.9% purity results covers the purity side.

Why two labs can differ on the same batch

A lab’s ± covers the scatter inside its own method, not a different basis or reference standard at another lab. Three published same-batch cases:

Published case	Product	mg results by lab
Peptide Critic, May 2026	Retatrutide, 10 mg kit	Vanguard 10.18 and 10.30; Analytical Formulations (AFI) 10.71 and 10.85; Janoshik 11.03 to 11.49; TrustPointe 12.296 and 12.328; Freedom Diagnostics 12.47 and 12.80
graymarket, September 2025	Retatrutide, one batch	Janoshik about 30; Chromate 30; Krause 26 on both vials
Krysia’s Substack, August 2026	Tirzepatide, 60 mg label	Chromate 71.63 and 70.99; Accumark 73.72 and 75.44; BTLabs 62.3

In the Peptide Critic case, each lab’s own results sat within about 4% of each other, while the labs spread by about 26%. Each method looked repeatable; the methods disagreed. Each case is a single published comparison, not proof that any lab is wrong. In the August 2026 piece, Janoshik said it standardises its own reference standards by CHNS and NMR and reports net peptide. Vanguard said it buys certified standards and identifies the counter-ion by ion chromatography. BTLabs said that labs use different standards.

Vials differ too. Krause’s answer notes that its margin excludes vial-to-vial variation, because routine tests use one vial, and Chromate attributes that variation to the manufacturer’s fill precision. Several vials show the spread, as our guide on how many vials to send explains. Finnrick says it estimates each lab’s offset from overlapping tests and its own anchor samples, but in September 2026 it had not published the offsets. Our guides on why labs disagree and same-batch comparisons go further.

Reading a result near the label or a limit

Uncertainty matters most when a result sits near a limit. ISO/IEC 17025 asks a lab that states pass or fail to apply a documented decision rule that takes uncertainty into account. ILAC G8 describes simple and guarded acceptance. Under simple acceptance, a result inside the limit passes. Under guarded acceptance, the limit moves inward by a guard band, often equal to U.

Take a client-chosen specification of 90 to 110% of a 10 mg label, and an illustrative result of 9.1 ± 0.4 mg (k = 2). Under simple acceptance it passes, because 9.1 is above 9.0. Under guarded acceptance with a 0.4 mg band, it needs 9.4 mg, so it fails. The range of 8.7 to 9.5 mg straddles the limit, and a bare “PASS” hides that.

Before you rely on a number, check that the report states:

1. the result with its ± and its k, and how the uncertainty was estimated;
2. how many preparations, injections and vials stand behind the result;
3. the reference standard: its source, lot and assigned content;
4. the basis of the mg: free base, salt or total powder;
5. digits that match the uncertainty;
6. the specification and decision rule behind any verdict.

How HK Analytical states accuracy

Every measured value on our reports carries its expanded uncertainty (k = 2), or the reason none is stated, and we round to match it. Content per vial is the mg of the target peptide as free base, measured against a reference standard, with the % of the label claim. If no qualified reference standard exists, the value is marked “estimated”. A verdict appears only with a stated specification and decision rule, as our quality page explains. Vial-to-vial consistency reports each vial’s mg with the mean, range and relative standard deviation. Every number comes with its raw data in a signed record; the example report shows the layout. We do not hold ISO/IEC 17025 accreditation.