An analytical method is used by people on instruments that do not reproduce every condition perfectly. A robustness study asks whether small, deliberate changes within realistic use conditions alter the performance that matters. Its value lies in identifying which settings need close control and which can vary without undermining the result.
Define what must remain reliable
ICH Q14 describes robustness in terms of maintaining expected performance under deliberate changes to analytical parameters. It connects this work with development knowledge, risk assessment and the resulting analytical control strategy.ICH — Analytical procedure development Q14 (opens in a new tab)
For a peptide impurity method, the critical outcome might be separation of a particular impurity from the main peak. For a content assay, bias and precision over the reporting range may be central. A study should name the outcome before choosing convenient measurements.
If the main peak area remains stable but an impurity merges into it, an apparently reassuring number can conceal reduced discrimination. Likewise, a consistent retention time does not demonstrate that a preparation still recovers the same amount.
Select plausible changes with a reason
ICH Q14 includes the duration of the procedure and the stability of sample preparations and reagents among the considerations for robustness. The selected factors should reflect knowledge of the procedure and its intended use.ICH — Analytical procedure development Q14 (opens in a new tab)
| Planned variation | Question to investigate |
|---|---|
| Prepared-sample waiting time | Does the reported composition change before analysis? |
| Mobile-phase preparation variation | Does the critical pair remain distinguishable? |
| A permitted equipment setting | Does the result remain suitable across its allowed range? |
| A routine preparation interval | Does changing the interval alter the recovered result? |
These are planning examples, not recommended operating conditions. The appropriate sizes of changes depend on the actual procedure. An extreme stress experiment can investigate a different question from routine robustness.
Record factors that were considered but excluded and the reason. Otherwise, an omitted risk can look indistinguishable from a demonstrated lack of effect.
Do not assume separate changes stay separate
Consider this original hypothetical response: the usual settings give 99.8 units; changing factor A alone gives 99.9; changing B alone also gives 99.9; changing both gives 101.5. The combined response is not predicted by simply adding the two individual changes.
The example illustrates an interaction, not a peptide specification or an acceptance decision. It shows why checking each factor separately can leave a combination unexplored. A suitable experimental design can investigate selected combinations while preserving a clear comparison with the usual conditions.
Replication helps distinguish a real factor effect from ordinary measurement variation. Randomising the order, where feasible, also reduces the risk that a gradual instrument change is mistaken for the effect of the planned factor.
The useful conclusion identifies the conditions actually supported. It should not silently turn two separately tested ranges into permission to combine every value within both ranges.
Turn findings into practical controls
Bongers and colleagues reported validation work on an antibody tryptic map that uncovered preparation-related side products and changes during extended autosampler storage. This was a particular protein-mapping study, but it demonstrates how procedural history can become part of the analytical result.Bongers and colleagues — Validation of a peptide mapping method for an antibody (opens in a new tab)
A finding may justify narrowing an allowed interval, clarifying a preparation instruction or adding a check that detects the relevant failure. Repeating the study without changing the procedure or its controls would leave the same practical vulnerability.
A summary should identify the tested factors, ranges, response measures and unresolved limitations. “Robust method” is less informative than an explanation of which variations were challenged and what remained acceptable.
Sources and further detail
- ICH — Analytical procedure development Q14 (opens in a new tab)
Final 2023 guideline, section 5.1. Short summary of robustness and control-strategy principles; examples are original.
- Bongers and colleagues — Validation of a peptide mapping method for an antibody (opens in a new tab)
Primary 2000 study, DOI 10.1016/S0731-7085(99)00181-8. The preparation and storage observations concern that mapping procedure, not all peptides.
Sources checked 19 September 2026. Worked examples are illustrative unless a supplied report is explicitly identified. This article has not undergone independent scientific peer review.