Bracketing & Matrixing (ICH Q1D)
ICH Q1D lets a stability programme skip some strength/pack combinations or time points — but only where the design, the justification and the pull records can show the missing data was never needed.

On this page · 8 sections
- 1What ICH Q1D is for
- 2Bracketing: testing only the extremes
- 3Matrixing: testing a planned subset at each time point
- 4When a reduced design is, and is not, applicable
- 5Potential risk and data evaluation
- 6Protocol and sample-pull record implications
- 7Use in regulatory submissions and supplements
- 8Status of the ICH Q1 consolidation (as of October 2026)
How does Bracketing & Matrixing (ICH Q1D) apply to your shop floor?
Pick your industry and scale — Ask V5 rewrites the definition in your context, gives a worked example, and shows what V5 does on day one.
01What ICH Q1D is for
ICH Q1D, Bracketing and Matrixing Designs for Stability Testing of New Drug Substances and Products (Step 4, 7 February 2002), is a short companion guideline to the parent ICH Q1A(R2) stability guideline. Q1A(R2) states that bracketing and matrixing can be applied to a stability study "if justified," but gives no detail on how; Q1D exists to supply that detail. It defines the two reduced-design techniques, states when each is and is not applicable, works through illustrative design tables, and sets minimum data-evaluation expectations for a reduced design.
A full stability design tests every combination of the study's design factors — typically batch, strength, and container size and/or fill — at every scheduled time point. A reduced design deliberately tests fewer than all of those combinations at some or all time points, on the documented assumption that the untested combinations behave the same as (or no better than) the tested ones. Q1D's own framing is that any reduced design "should have the ability to adequately predict the retest period or shelf life," and that the potential risk of landing on a shorter shelf life than a full design would have supported should be considered before a reduced design is chosen.
Q1D does not apply on its own; it is read together with Q1A(R2) for study conditions and time points, and with ICH Q1E for how the resulting (possibly incomplete) dataset is evaluated to justify a retest period or shelf life. A firm using bracketing or matrixing is still running a Q1A(R2)-structured study — it is only the matrix of which samples get pulled and tested at which time point that changes.
02Bracketing: testing only the extremes
Q1D defines bracketing as "the design of a stability schedule such that only samples on the extremes of certain design factors (e.g., strength, container size and/or fill) are tested at all time points as in a full design." Every time point that would appear in a full study still appears for the batches at the extremes; what is omitted is testing of the intermediate strengths or pack sizes at any time point. The design rests on the assumption that if the two extremes are stable, everything in between is at least as stable.
That assumption has to be demonstrable, not asserted. Q1D states plainly that a bracketing design "would not be considered appropriate if it cannot be demonstrated that the strengths or container sizes and/or fills selected for testing are indeed the extremes." For strength, bracketing fits cleanly when the strengths come from the same formulation platform — the guideline's own examples are capsules of different strengths filled from the same powder blend with different plug sizes, tablets compressed in different weights from the same granulation, and oral solutions that differ only in minor excipients such as colourants or flavourings. Where the relative amounts of drug substance and excipient actually change across strengths, bracketing can still be used but needs justification, for example comparable stability profiles demonstrated on clinical or development batches; where different strengths use different excipients altogether, Q1D says bracketing "generally should not be applied."
For container closure systems, bracketing works where container size or fill volume varies while everything else about the container stays constant. If both size and fill vary together, the guideline warns against assuming the largest and smallest containers are automatically the stability extremes — the comparison has to look at container wall thickness, closure geometry, surface-area-to-volume ratio, headspace-to-volume ratio, and water-vapour or oxygen permeation rate per dosage unit or unit fill volume. Bracketing across different closures on the same container is possible with justification, such as a comparison of relative permeation rates.
Q1D's own illustrative example (its Table 1) is a product sold in three strengths (50 mg, 75 mg, 100 mg) and three container sizes (15 ml, 100 ml, 500 ml). The bracketing design tests the 15 ml and 500 ml containers at every strength and every time point, on the basis that those two sizes are the demonstrated extremes; the 100 ml container is not tested at all. If, after the study starts, one of the bracketed extremes is dropped from the marketed range, Q1D allows the study to continue supporting the remaining intermediates, provided the firm commits to running a post-approval stability study on whatever extreme actually ends up marketed. And if the two tested extremes turn out to differ in stability, the guideline requires treating the intermediates as no more stable than the less stable extreme — the shelf life assigned to the whole bracket is capped by the worse-performing endpoint, not averaged.
03Matrixing: testing a planned subset at each time point
Matrixing is defined differently: "the design of a stability schedule such that a selected subset of the total number of possible samples for all factor combinations would be tested at a specified time point. At a subsequent time point, another subset of samples for all factor combinations would be tested." Rather than dropping whole strengths or pack sizes, matrixing tests every combination eventually, just not at every time point — different subsets rotate through the schedule. Q1D is explicit that "differences in the samples for the same drug product should be identified as, for example, covering different batches, different strengths, different sizes of the same container closure system, and possibly, in some cases, different container closure systems," and that matrixing can extend to a secondary packaging system where that packaging contributes to product stability.
Each storage condition — long-term, accelerated, intermediate — is matrixed separately under its own design, and matrixing should not be performed across different test attributes; a design where assay is matrixed one way and dissolution another requires its own justification if attempted.
Q1D sets firm minimum coverage rules regardless of how aggressive the reduction is. All selected factor combinations must be tested at the initial and the final time point. If full long-term data for the proposed shelf life will not be available before approval, all selected combinations must also be tested at 12 months or the last time point before submission. At least three time points, including initial, must have data for each selected combination within the first 12 months of the study. For accelerated or intermediate conditions, each selected combination needs a minimum of three time points, including initial and final. These floor requirements are why Q1D's own worked examples describe reductions that fall short of a clean half or third: a nominal "one-half reduction" design that would eliminate every other time point actually removes 15 of 48 combinations once initial, 12-month and final full testing are restored, and a nominal "one-third reduction" removes only 10 of 48 for the same reason.
Q1D's own illustrative tables work through a two-strength product matrixed on time points only (its Table 2, with "one-half reduction" and "one-third reduction" variants), and a three-strength, three-container-size product matrixed either on time points alone (Table 3a, where every batch/strength/container combination is tested, just not at every time point) or on time points and factors together (Table 3b, where some factor combinations are skipped entirely at some time points). The guideline notes these are illustrative only — "should not be considered the only, or the most appropriate, designs in all cases" — not a menu to copy without adaptation to the actual product and portfolio.
04When a reduced design is, and is not, applicable
Q1D states that reduced designs can be applied to the formal stability study of most drug product types, with additional justification needed for complex drug delivery systems that have significant drug-device interaction potential. For drug substances specifically, the guideline is blunt: "matrixing is of limited utility and bracketing is generally not applicable." Reduced designs in Q1D are primarily a drug-product tool, aimed at multi-strength and multi-pack product lines, not a general technique for substance stability.
Applicability of matrixing is explicitly tied to data variability. Q1D lists the factors to weigh — knowledge of data variability, the expected stability of the product, the availability of supporting data, stability differences within or among factors, and the number of factor combinations in the study — and then states the decision rule directly: matrixing is appropriate where supporting data show only small variability; where supporting data show moderate variability, a matrixing design must be statistically justified (for example, an evaluation of the proposed design's power to detect differences in degradation rates, or its precision in shelf-life estimation); and "if the supportive data show large variability, a matrixing design should not be applied" at all. The degree of reduction a firm can propose scales with the number of factors and factor levels involved — more strengths and pack configurations support a larger reduction — but Q1D ties this back to the same constraint as bracketing: whatever is proposed still has to support a reliable shelf-life estimate.
Q1D also flags that bracketing and matrixing rest on different principles, so combining them in one design needs its own careful justification rather than being treated as simply stacking two cost-reduction techniques. And it allows switching mid-study: a firm can move from a reduced design to full testing, or to a less reduced design, if justified, but the statistical analysis has to be adjusted for the resulting change in sample size, and once changed, the new testing level continues for the remaining time points — a firm cannot partially step back toward full testing for one time point and revert.
05Potential risk and data evaluation
Q1D is direct about the statistical cost of a reduced design: matrixing on factors other than time points "generally has less precision in shelf life estimation and yields a shorter shelf life than the corresponding full design," and may lack the power to detect certain main or interaction effects — which can lead to data from different design factors being incorrectly pooled during shelf-life estimation. If the reduction in tested factor combinations is excessive and the data cannot be pooled to support a single shelf life, the guideline warns it may become impossible to estimate shelf life for the combinations that were never tested at all. By contrast, a design that matrixes only on time points (testing every factor combination, just not at every time point) is described as usually retaining an ability similar to a full design to detect rate differences among factors and establish a reliable shelf life, because linearity is assumed and full testing is still performed at the initial and final time points.
On data evaluation, Q1D's own instruction is one sentence: "Stability data from studies in a reduced design should be treated in the same manner as data from full design studies." The statistical treatment described in ICH Q1E for combining long-term and accelerated data, assessing whether batches can be pooled, and extrapolating a shelf life applies to a bracketed or matrixed dataset the same way it applies to a full dataset — Q1D does not create a separate, looser evaluation standard for reduced designs. What differs is only the data available to feed into that evaluation, and the risk discussion above is Q1D's way of flagging that a thinner dataset produces a less precise, often shorter, conclusion.
06Protocol and sample-pull record implications
Because bracketing and matrixing change which samples get pulled at which time point, the stability protocol itself has to spell out the design explicitly: which strengths and container configurations are included in the study, which specific combinations are tested at which time points, the justification for the bracket or matrix chosen (including the supporting data relied on for variability and comparability), and what happens if a bracketed extreme or matrixed combination is dropped from the marketed range mid-study. A protocol that says "bracketing applied" without naming the exact combinations and time points gives a reviewer nothing to check the pull schedule against.
The sample-pull and testing records generated during execution need to trace back to that protocol design, not to a generic pull calendar. For a bracketed study, the pull record should show that the tested strengths or container sizes are in fact the ones designated as extremes, at every time point in the protocol — a missed pull on an extreme is not recoverable by substituting an intermediate, since the intermediate was never validated as representative. For a matrixed study, the pull record needs to show which specific subset of combinations was due at each time point per the matrix design, and that the mandatory full-coverage time points (initial, 12-month or last-pre-submission, and final) were in fact tested for every combination, not just the subset that happened to be convenient. A reviewer reconstructing a shelf-life justification years later needs to be able to see, from the record alone, that the minimum coverage rules in Q1D section 2.4.2 were met for every factor combination in the matrix, not inferred from the overall sample count.
In practice this means the record of a reduced-design stability study is only as useful as the link between three things: the approved protocol stating the design and its justification, the pull/test schedule actually executed, and the disposition of any mid-study change (added batches, dropped configurations, a shift from reduced to full testing) with the date it took effect. A platform used to manage stability pulls can hold the matrix or bracket as the schedule definition and flag a missed or substituted pull against it, with the quality unit still making the disposition call on what a gap means for the study — the software does not decide whether a gap is acceptable, it only makes the gap visible against the documented design. sample sampling instructions dossier extract and sample test methods dossier extract show the kind of document extracts a reduced-design protocol and its supporting method information would sit alongside in a dossier.
07Use in regulatory submissions and supplements
Bracketing and matrixing are design choices made for the initial stability programme supporting a marketing application, and the same logic applies when stability support is needed for a post-approval change — for example adding a new strength, pack size, or container to an already-approved product line. Whether a reduced design is acceptable to support such a change, and what data a regulator expects before or after approval, is governed by the applicable post-approval change guidance and regulation for the product and region in question (in the United States, for drugs, the relevant CBE/PAS/annual-reportable framework and FDA's SUPAC and stability guidances), not by Q1D itself — Q1D only addresses how the stability study is designed once a change in scope has been decided.
Q1D's own commitment language is narrow and specific: it applies to the case where a bracketed extreme stops being marketed mid-study, in which case the firm is expected to commit to post-approval stability studies on whichever extreme is actually marketed. Beyond that single scenario, Q1D does not describe a general framework for supplement or variation filings; the governing post-approval change guidance for a given jurisdiction determines what stability data package, if any, a bracketing- or matrixing-based study needs to support before a change can be implemented.
08Status of the ICH Q1 consolidation (as of October 2026)
ICH is consolidating the Q1A(R2), Q1B, Q1C, Q1D, Q1E, and Q5C stability guideline series into a single revised guideline, titled ICH Q1 "Stability Testing of Drug Substances and Drug Products," together with related revisions to Q5C. Work began with a concept paper and business plan endorsed in November 2022. The consolidated draft reached Step 2 (consensus draft text endorsed by the ICH Assembly) on 11 April 2025 and was opened for a Step 3 public consultation period running from April to September 2025; FDA issued its own draft guidance mirroring the ICH Step 2 text in June 2025, describing it explicitly as "a consolidated revision of the ICH Q1A(R2), Q1B, Q1C, Q1D, Q1E, and Q5C series," and noting it adds stability guidance for product categories such as advanced therapy medicinal products, vaccines and other complex biological products not previously covered by the existing series FDA draft Q1 guidance.
As of the most recent published ICH Q1 Expert Working Group work plan (dated 11 February 2026), the guideline remains at Step 2/Step 3: the expert working group is still resolving comments from the 2025 public consultation, with a plenary working-party engagement targeted for September 2026 ahead of an anticipated Step 3 sign-off and Step 4 adoption in November 2026, and delivery of training material projected for June 2027 ICH Q1 EWG work plan, Feb 2026. That means the standalone ICH Q1A(R2), Q1D and Q1E guidelines described in this article remain the current, adopted Step 4 texts as of October 2026; the consolidated Q1 has not been finalized or adopted, and this article should be revisited once Step 4 adoption occurs and the content of the final Q1 text — including whatever it specifies about bracketing and matrixing — is confirmed against the adopted text rather than the 2025 draft.
Frequently asked questions
Q.Can bracketing and matrixing be used together in the same stability study?+
Q1D allows it but calls for careful, specific justification, stating that the two techniques are reduced designs based on different principles and that combining them in one design needs scientific justification beyond what would support either technique alone.
Q.Does matrixing mean some strengths or pack sizes are never tested?+
No — that describes bracketing. In matrixing, every factor combination is eventually tested; what is reduced is the number of time points at which each combination is tested, with mandatory full testing of every combination still required at the initial, 12-month (or last pre-submission), and final time points.
Q.Is bracketing or matrixing generally used for drug substance stability studies?+
Q1D states that for drug substances, matrixing is of limited utility and bracketing is generally not applicable; the techniques are aimed primarily at drug product studies with multiple strengths, container sizes or fills.
Q.What happens if the two bracketed extremes turn out to have different stability?+
Q1D requires treating the intermediate strengths or pack sizes as no more stable than the less stable extreme, so the shelf life assigned across the bracket cannot exceed what the least stable tested extreme supports.
Q.Has ICH replaced Q1D with a new consolidated Q1 guideline?+
Not yet, as of October 2026. The consolidated ICH Q1 guideline (merging Q1A(R2), Q1B, Q1C, Q1D, Q1E and Q5C) reached Step 2 in April 2025 and underwent Step 3 public consultation through September 2025, with Step 4 adoption anticipated around November 2026 per the ICH Q1 work plan dated February 2026. Until Step 4 adoption is confirmed, the standalone Q1D guideline described here remains the current, adopted text.
Primary sources
- ICH Q1D, Bracketing and Matrixing Designs for Stability Testing of New Drug Substances and Products, Step 4 (7 Feb 2002) — reviewed 9 Oct 2026
- FDA draft guidance, Q1 Stability Testing of Drug Substances and Drug Products (June 2025) — reviewed 9 Oct 2026
- ICH Q1 EWG Work Plan, 11 February 2026 — reviewed 9 Oct 2026
Further reading
- ICH Q1A(R2) stability storage conditionsThe parent guideline that bracketing and matrixing designs are applied within.
- Stability ProgramHow a stability program is structured end to end, including protocol and pull scheduling.
- Stability-Indicating MethodThe analytical method foundation that any stability design, reduced or full, depends on.
- Supplement Stability and Shelf LifeHow Q1A(R2)-aligned stability practice is applied in dietary supplement shelf-life support.
- ICH Stability Zone IVbClimatic-zone storage conditions that bracketing/matrixing designs are run against.
- Sample sampling instructions dossier extract
- Sample test methods dossier extract
Want to see how Bracketing & Matrixing (ICH Q1D) could fit into your own records and workflows? Explore the related V5 pages or talk to our team about what applies to your operation.
