Precision: Repeatability, Intermediate Precision and Reproducibility

Two laboratories both report a relative standard deviation under 2%. One ran six injections of one solution in an afternoon. The other ran the full procedure across three analysts, two instruments and four weeks. Both figures are honest. They are not the same measurement, and ICH Q2(R2) gives them different names for exactly that reason.

Quick answer

What is precision? ICH Q2(R2) defines it as the closeness of agreement — the degree of scatter — between a series of measurements obtained from multiple samplings of the same homogeneous sample under the prescribed conditions, usually expressed as the variance, standard deviation or coefficient of variation of a series of measurements. The guideline states plainly that precision can be considered at three levels: repeatability, intermediate precision and reproducibility.

The three differ in what is allowed to vary between measurements, and that is the entire distinction.

The three levels

The exact minimums the guideline sets

Repeatability is unusual among validation characteristics in having a hard number attached. ICH Q2(R2) says it should be assessed using either:

• a minimum of 9 determinations covering the reportable range — for example 3 concentrations with 3 replicates each; or
• a minimum of 6 determinations at 100% of the test concentration.

A determination, in the guideline's own glossary, is the reported value from single or replicate measurements of a single sample preparation as per the validation protocol. This is the same principle that governs accuracy, covered in accuracy and recovery: what is being counted is preparations, not injections.

For intermediate precision the guideline sets no fixed number, because the extent depends on the circumstances under which the procedure is intended to be used. The applicant should establish the effects of random events on precision, studying typical variations including different days, environmental conditions, analysts and equipment, as relevant. Two further points are easy to miss: studying these effects individually is not necessary, and the use of design of experiments studies is encouraged.

Reproducibility is the one almost nobody does

ICH Q2(R2) is direct about this. Reproducibility is assessed by means of an inter-laboratory trial, and investigation of reproducibility is usually not required for regulatory submission. It should be considered in cases of standardisation of an analytical procedure — for instance for inclusion of procedures in pharmacopoeias — and where procedures are conducted at multiple sites.

Which explains why published inter-laboratory peptide data is rare enough to be notable. The United States Pharmacopeia study on oxytocin described in how much quantification methods agree is reproducibility in the strict sense: national metrology and regulatory institutes across several countries, one material, three methods. That study exists precisely because pharmacopoeial standardisation is one of the cases where the guideline says reproducibility should be considered.

Why the level changes what a number is worth

Scatter accumulates as more is allowed to vary. A repeatability figure is the floor — the best the method will ever look. Intermediate precision is larger, and it is the honest number for judging whether a difference you observed is real. Reproducibility is larger still, and it is the only one that speaks to agreement across sources.

This is why comparing a supplier's tight precision figure against your own bench scatter is not a like-for-like comparison, and why the advice in lot-to-lot variability and bridging is to compare within a supplier rather than across suppliers.

What gets reported

The guideline asks for the standard deviation, the relative standard deviation (coefficient of variation), and an appropriate 100(1−&alpha)% confidence interval or justified alternative statistical interval, with the observed interval compatible with the precision acceptance criteria.

A relative standard deviation alone is therefore an incomplete report by the guideline's own standard — the interval is part of the result, as it is for accuracy and for the certified values discussed in measurement uncertainty.

Combining accuracy and precision

ICH Q2(R2) permits an alternative to evaluating the two separately: considering their total impact by assessing against a combined performance criterion, evaluated using a prediction interval, a tolerance interval or a confidence interval.

The reasoning is practical. A user of a result does not experience bias and scatter separately; they experience how far the reported number might be from the truth. A combined criterion measures that directly.

What a research buyer can build from this

• Measure your own intermediate precision, not your repeatability. Six preparations on one afternoon will flatter you. Spread them across days and, if possible, people.
• Use that figure as your detection threshold. Differences smaller than your own scatter are not findings. This single rule prevents most false lot-to-lot conclusions.
• Record which level any number represents. A precision figure without its conditions cannot be interpreted later, by you or anyone else.
• Vary things together, not one at a time. The guideline explicitly says studying the effects individually is not necessary — which makes a realistic estimate far cheaper than it first appears.
• Do not expect reproducibility figures to exist. They are rare by design. Their absence is normal, not a red flag.

Key takeaways

• ICH Q2(R2) defines precision at three levels: repeatability, intermediate precision and reproducibility.
• Repeatability requires a minimum of 9 determinations across the reportable range, or a minimum of 6 at 100% of the test concentration.
• Intermediate precision covers days, environmental conditions, analysts and equipment within one laboratory; the effects need not be studied individually, and design of experiments is encouraged.
• Reproducibility is an inter-laboratory trial and is usually not required for regulatory submission.
• Pharmacopoeial standardisation is one of the cases where reproducibility should be considered — which is why multi-laboratory peptide studies exist at all.
• Reporting should include standard deviation, relative standard deviation and a confidence interval, not an RSD alone.
• Accuracy and precision may be assessed together against a combined performance criterion using a prediction, tolerance or confidence interval.

Frequently asked questions

Is a 1% RSD a good result?

Unanswerable without knowing the level. One percent repeatability on replicate injections is unremarkable; one percent intermediate precision across analysts and weeks would be excellent. The conditions carry the meaning.

How many replicates should I run on my own material?

Six at the working concentration matches the guideline's lower option and is achievable. Spreading those six across different days converts a repeatability estimate into something closer to intermediate precision, which is the more useful number.

Does good precision mean the result is correct?

No. A method with a systematic loss repeats that loss faithfully. Correctness is a separate characteristic — see accuracy and recovery.

Why does the guideline allow effects to be varied together?

Because the goal is an estimate of how much the result moves in ordinary use, not an attribution of the movement to a particular cause. Separating causes is a development question, which is why design of experiments is encouraged rather than required.

Should I ask a supplier for precision data?

It is a reasonable question, and the informative part of the answer is which level it refers to. A figure described simply as precision, with no conditions attached, tells you less than it appears to.

Sources

• International Council for Harmonisation. ICH Harmonised Guideline Q2(R2): Validation of Analytical Procedures. Final version adopted 1 November 2023. ICH guideline
• Li C, Bhavaraju S, Thibeault MP, Melanson J, Blomgren A, Rundlöf T, et al. Survey of peptide quantification methods and comparison of their reproducibility: a case study using oxytocin. J Pharm Biomed Anal. 2019;166:105–112. DOI

Journal citation retrieved via PubMed. The guideline text is quoted from the adopted ICH Q2(R2) document.

This article describes analytical methodology only. It contains no usage instructions, no dosing information and no protocols of any kind.

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