Purified Water vs WFI: How to Select the Appropriate Water Quality for Each Pharmaceutical Use

During an inspection, the question that opens the water chapter is almost never “what water grade do you use?”. It is “why that grade, for that use, and where is it written down?”. The first question is closed by a specification; the second requires a documented rationale linking every point of use to the assigned water quality, to the source that justifies it, and to the evidence that the quality is actually delivered at that point.

In most sites the specification exists; the rationale does not. It lives in the memory of whoever designed the system ten years earlier. The result is two symmetrical and equally expensive defects. Over-specification, with WFI distributed everywhere “to be safe”, multiplies utility consumption, sanitisable surfaces, sampling points and deviations to investigate. Under-specification uses Purified Water at a step where the residue ends up in the finished product, with nobody having assessed the contribution in bioburden and endotoxins.

This article deals with the selection criterion, not with generation technology: how to build and maintain a use → water grade → rationale → documentary evidence matrix that stands up to inspection in Europe and in the United States.

The problem is not choosing: it is demonstrating why

Selecting a water grade is a quality decision, not an engineering one. The recurring mistake is reversing the order: you look at what the installed system can deliver and write the specification accordingly. That is the opposite of the expected path, and it produces documentation that describes the status quo instead of justifying it.

The correct path starts from the use. For each point you establish what the water does: it enters the product, it touches the product, it touches a surface that will touch the product, or it stays in a utility with no contact. The minimum grade follows from that classification, the internal specifications follow from the grade, and monitoring, sampling and release criteria follow from the specifications. If the chain is not written down, every link is an unsupported statement.

The same logic governs the drafting of URS for a water system: the URS should not be written before the use matrix, because it is the matrix that defines how many loops are needed, at what grade and with which points of use.

What a monograph defines, and what it does not

The most widespread misunderstanding is believing that a pharmacopoeial monograph tells you when to use that water grade. It does not: it defines what that water must be, not where it is to be used. The use → grade mapping sits elsewhere, and in the EU and the US it sits in different places.

USP: two short definitions, two controlled attributes

The USP Purified Water monograph reads: “Purified Water is water obtained by a suitable process. It is prepared from water complying with the U.S. EPA National Primary Drinking Water Regulations or with the drinking water regulations of the European Union or of Japan, or with the WHO Guidelines for Drinking Water Quality. It contains no added substance.”

The USP Water for Injection monograph reads: “Water for Injection is water purified by distillation or a purification process that is equivalent or superior to distillation in the removal of chemicals and microorganisms.”

Neither definition mentions the use: they anchor the grade to the feed water and to the purification process. Anyone looking in those lines for permission to use PW in a parenteral will not find it, in either direction.

The attributes actually controlled by the two monographs are conductivity, under USP <645>, and total organic carbon, under USP <643>. Chapter <645> has a sequential three-stage structure, of which Stage 2 is the offline route and Stage 3 adds a neutral electrolyte (KCl) to raise ionic strength and measure pH accurately; the applicable numerical values must be read in the current chapter and are not reproduced here. Chapter <643> works on the difference between the sample response and the reagent water response, and states that it does not set how often the system suitability test should be run: the frequency is left to the user's risk assessment.

What really changes the way a specification is written, however, is something else, stated by USP directly: “Because of the various uses of these waters, microbial requirements are not included in these monographs.” pH, nitrates and heavy metals are no longer USP requirements for these waters. The consequence is significant: “USP-compliant water” is not a microbiological specification, and a document limited to that phrase has defined no microbiological criterion at all.

Ph. Eur.: separate monographs and one suppressed chapter

On the European side the structure is different. The relevant monographs are 0008 Water, purified, 0169 Water for injections and 2249 Water for preparation of extracts, the last one separate and tied to a specific use: its very existence shows that the European compendial logic is organised by intended use, not by a generic purity ladder.

Monograph 0169 was revised by a decision adopted at the 154th session of the European Pharmacopoeia Commission (March 2016), effective 1 April 2017, to allow WFI production by a purification process equivalent to distillation. A change in the architecture of grades followed: monograph 1927 Water, highly purified was suppressed by a decision of the 160th session (March 2018), effective 1 April 2019, having been made redundant precisely by the revision of 0169.

The logic is worth understanding, because HPW keeps appearing in the documentation of sites that never took the change on board. It was an intermediate grade: quality comparable to WFI but obtainable by non-distillation routes, for uses where WFI was not mandated. Once processes equivalent to distillation were allowed, that grade lost its purpose. Today the Ph. Eur. compendial ladder relevant to this decision is PW and WFI, and a specification issued after 1 April 2019 that assigns “Highly Purified Water” to a point of use, with no transition note, is a documentation defect.

Chemical attributes are governed by general chapters 2.2.38 Conductivity and 2.2.44 Total organic carbon in water for pharmaceutical use. Recombinant factor C (rFC) is permitted for the bacterial endotoxin test in the water monographs, by a decision of the 175th session (March 2023), Suppl. 11.4, implementation from 1 April 2024. The numerical limits of the monographs, including the endotoxin limit in 0169, must be read in the current edition and are not reproduced here.

Monographs move, and almost no matrix anticipates it. In the edition in force in August 2026 — 12th Ed., Issue 12.3, applicable from 1 July 2026 — the TOC test replaced the oxidisable substances test in sterilised water for injections, with revision of 0169, 0008 and 2.2.44; 0169 and 0008 were further revised in Issue 13.1, with implementation from 1 January 2027. A matrix that cites monographs without regulatory watch and without change control ages by itself.

A compendial requirement and a GMP requirement are not the same thing

The compendial requirement is the set of monograph attributes: it states what that water must be in order to be called Purified Water or Water for Injection. The GMP requirement for system control concerns instead the ability of the installation to deliver that water reproducibly, and it lives in the GMPs, not in the pharmacopoeia. Annex 1 of EudraLex Volume 4, in operation since 25 August 2023, places water quality among the elements of the Contamination Control Strategy at clause 2.5(v); at clause 6.10 it requires WFI to be produced from water conforming to specifications defined during qualification; at clause 6.15 it requires WFI systems to include continuous monitoring such as TOC and conductivity.

None of these requirements appears in a monograph: they are system obligations triggered by the grade selected. Choosing WFI for a point of use does not merely raise a specification, it imports a package of design, qualification and continuous TOC and conductivity monitoring obligations to be funded for the whole life of the installation. This is why over-specification is not a neutral, prudent choice: it is a permanent compliance cost.

On the microbiological side the separation is even sharper. USP chapter <1231> Water for Pharmaceutical Purposes, official since 1 December 2021, is an informational chapter: numbering above 1000 makes it non-binding. It indicates action levels — 100 cfu/mL for Purified Water and 10 cfu/100 mL for Water for Injection — described as levels above which the water is unfit for use and which trigger an investigation. They are not compendial limits: USP asks users to establish in-house specifications or fitness for use microbial levels. Citing them as “USP limits” is an error in qualifying the source, and it does get picked up; reporting them as internal action levels, approved after comparison with the site's own qualification and trend data, is correct. The figure may coincide, the rationale does not, and it is the rationale that is asked for. The subject is developed in the article on microbiological and endotoxin control in PW and WFI.

Intended use: the primary criterion under the EMA guideline

Guideline EMA/CHMP/CVMP/QWP/496873/2018 Guideline on the quality of water for pharmaceutical use, in force since 1 February 2021 and replacing CPMP/QWP/158/01 of 2002, is the document that in Europe explicitly links the use to the minimum grade. It is the source to cite in the “rationale” column of the matrix, with a precise section reference.

Section 5.1 — water as an excipient

Section 5.1 contains a table indicating the minimum water grade where water is an excipient of the product, differentiating between sterile and non-sterile products and by route of administration. Its detailed content is not reproduced here: it must be read in the guideline and carried into the matrix with the exact citation.

What matters methodologically is the structure of the criterion: the grade does not depend on batch size, on the department or on the installed system, but on two variables that belong to the dossier and not to engineering. A matrix that assigns a grade without stating, for each row, the product, whether it is sterile or non-sterile, and the route of administration is not verifiable, because the inputs to the decision are missing.

The guideline also defines a minimum grade: using a higher grade is allowed; what is not allowed is doing so unknowingly, because the higher grade drags along system obligations and costs that somebody will have to carry.

Section 5.2 — active substances and dosage forms

Section 5.2 addresses water used in the manufacture of active substances and dosage forms and recognises a principle that 5.1 does not cover: in the early stages of synthesis, lower water grades are acceptable. It is the formal acknowledgement that risk is not uniform along the process.

In the matrix, therefore, the “stage” column is not decorative. Two draw-off points feeding the same reactor may legitimately carry different grades if one serves an early stage of synthesis and the other a stage downstream of which no step remains capable of removing the contribution of the water. The deciding question is always the same: between this point and the finished product, what removes what the water introduces? If the answer is “nothing”, the grade must be aligned with that of water used as an excipient.

Anyone adopting WFI produced by a non-distillation route must consider one further implication: the revision of 0169 reflected in section 4.2 allows single or double pass RO coupled with EDI, ultrafiltration or nanofiltration, but requires prior notification to the supervisory GMP authority before introducing RO. That is why the grade decision must be shared with Regulatory Affairs before it reaches the engineering department. The comparison between the two routes is covered in the article on membrane WFI generation versus distillation.

The operational roles of water: four different questions

Water as an excipient

This is the simplest case, because the guideline answers directly. The water enters the formulation, remains in the finished product, contributes to the impurity profile and, for parenterals, to the endotoxin load. The grade is read from the section 5.1 table and the row closes with a reference to the dossier and to the product specification.

Water as a process aid

Here the decision is less automatic. The water is not declared as a component of the formulation but it touches the product or the intermediate: granulation water later removed by drying, water for buffer solutions, for rehydrating an intermediate, for transfer operations. The formulation does not contain it, but its non-volatile residues and its endotoxins remain.

The operational criterion is the residue: you assess what part of the water survives the process. Endotoxins are not removed by drying; bioburden can be reduced by a downstream step, the products of microbial metabolism cannot. If no downstream step has demonstrated removal capability, the grade must be treated as that of excipient water, and the row must state which step the removal is attributed to and on what evidence. Otherwise “it gets removed later anyway” remains an unsupported statement.

Cleaning water

Wash water is not homogeneous: different phases of the same cycle tolerate different grades. Pre-wash and intermediate rinses remove gross residues and detergent and are not the last contact with the surface: the grade may be lower, provided the transition to the higher grade is defined, justified and verifiable within the cycle. The water used to prepare detergent and sanitising solutions must also be captured, since it often stays outside the matrix despite being in direct contact with product surfaces.

Final rinse

The final rinse is the only phase in which water is the last thing to touch the surface before that surface touches the next batch: whatever it leaves behind — non-volatile residue, bioburden, endotoxins — becomes part of the contamination profile of that batch.

It must be stated plainly that the regulatory sources cited in this article do not set a water grade for the final rinse. The criterion that follows is a widely adopted good engineering practice, to be supported by a site risk assessment and not by a regulatory citation: the final rinse water grade should be at least equal to the grade required for water in contact with the product manufactured on that equipment. The rationale must be supported by evidence — cleaning validation rinse samples, loop trend data, hold times between end of cycle and use, storage conditions of the cleaned equipment. Without those data the choice remains an opinion, whichever grade was selected.

Sterile and non-sterile products: what changes in the rationale

For non-sterile products the dominant question is the presence of objectionable microorganisms, that is, organisms relevant to that route of administration and to that patient: total count is an indicator, identification is the information. For sterile products a further attribute applies with no equivalent: bacterial endotoxins are not removed by terminal sterilisation or by sterilising filtration and cannot be recovered downstream. A system that sterilises correctly but starts from water with an uncontrolled endotoxin load produces batches that are sterile and non-conforming.

US GMPs place this distinction in 21 CFR 211.113: paragraph (a) requires written procedures to prevent contamination by objectionable microorganisms in non-sterile products, paragraph (b) procedures to prevent microbiological contamination in sterile products. There is no figure: there is an obligation to have a procedure and to make it effective. The figure is written by the site.

Regional differences: the divergence is about where the rule lives

A site supplying both markets does not manage two different purity ladders, but two different regulatory architectures. The table reports only the differences verified in the sources cited.

For markets referring to WHO, the applicable document is WHO TRS 1033, Annex 3 (2021) Good manufacturing practices: water for pharmaceutical use, which replaced TRS 970 Annex 2 of 2011: if the site is involved in prequalification or supplies markets that adopt it, the “reference” column of the matrix must cover it explicitly.

Building the use → grade → rationale → evidence matrix

The matrix is a controlled document, not a departmental spreadsheet: it has an owner, a review frequency, a link to change control and a demonstrable reconciliation with the P&ID and with the master batch records.

Step 1 — census

Start from the complete list of points of use, taken from the P&ID and reconciled with batch records and cleaning procedures. This is the step that fails most often, and always in the same way: the points that do not belong to production are missing. Detergent preparation, component washing, laboratory glassware washing, ancillary utilities, points installed for a discontinued product and never removed. A point that is not in the census is a point that is not specified and not sampled.

Step 2 — role classification

Each entry is classified as excipient, process aid, intermediate cleaning, final rinse, analytical use or non-contact utility. The classification must be justified in one line of text, not left implicit in the name of the point.

Step 3 — assigning the minimum grade

For entries covered by sections 5.1 or 5.2 the grade is read from the guideline and reported with the precise citation. For entries not covered — typically cleaning, final rinse and analytical use — the grade is determined through a structured risk assessment, consistent with the approach described in the article on quality risk management applied to water systems.

Step 4 — rationale and evidence

Each row closes with the two inspectable columns: the rationale, with source and precise reference, and the documentary evidence proving that the grade is actually delivered at that point. The evidence is not the specification: it is the qualification report, the monitoring record, the trend data, the cleaning validation report.

The layout below is the minimum format for a row. The grade and rationale columns are left to be completed: their content depends on reading the guideline and on the site's own data, and no generic table can substitute for it.

For entries the guideline does not cover, an explicit criterion is needed. The matrix below lists the criteria to be assessed; the weighting column is deliberately empty, because weighting depends on the product portfolio, on patient criticality and on the architecture of the installation, and must be assigned and approved by the site.

Worked example: Site Delta

Site Delta is a fictitious site, used here as an example. It manufactures a non-sterile oral solution and a parenteral lyophilisate, with a PW loop serving the whole plant and a WFI loop limited to the sterile area. The question arises in equipment washing: the final rinse for sterile area equipment is fed from the PW loop, following a decision taken at design stage and never formalised.

The census surfaces three points missing from every list: detergent solution preparation, a residual point from a discontinued product, and laboratory glassware washing. The first two enter the matrix, the third is classified as analytical use with its own criteria, and the residual point opens a change control for physical removal.

For excipient water the grade is read from the section 5.1 table: for the lyophilisate according to the parenteral route, for the oral solution according to the oral route and the non-sterile nature of the product. Both rows close without discussion.

The final rinse is not covered by the table. The team applies the criteria matrix, assigns weights according to its own portfolio and checks the decisive criterion: between the rinse residue and the lyophilisate there is no endotoxin removal step. The conclusion is to align the final rinse for equipment dedicated to the lyophilisate with the grade of the excipient water, extending the WFI loop to equipment washing as a change control with design and qualification impact.

The relevant point is not the conclusion, which at another site could be different and equally defensible. It is that the rationale now exists, cites its source, rests on cleaning validation and trend data, and is traceable to a change control. Before, there was only a practice.

Frequent mistakes

• Writing “USP-compliant water” as the only microbiological specification: USP monographs contain no microbiological requirements.
• Reporting 100 cfu/mL and 10 cfu/100 mL as “USP limits”: they are action levels from a non-binding informational chapter, which the user must translate into justified in-house specifications.
• Adopting WFI everywhere “to be prudent”, without assessing the system obligations and permanent costs it brings.
• Confusing the compendial requirement with the GMP requirement for system control, and treating monograph compliance as proof that the installation is under control.
• Failing to distinguish early stages of synthesis, for which section 5.2 allows lower grades, from final stages.
• Leaving out of the matrix the water used to prepare detergents and sanitisers and the water used for component washing.
• Assigning a grade to the final rinse without documenting either the criterion or the supporting evidence.

Inspection red flags

• Documents issued after 1 April 2019 assigning “Highly Purified Water” to a point of use, with no transition note following the suppression of monograph 1927.
• A use → grade matrix that is missing, or that cannot be reconciled with the P&ID and the point of use list.
• A specification declaring WFI at a point served by a loop qualified for PW.
• Alert and action levels identical for all points of a loop, with no reference to the qualification data for those points.
• A rationale resting on “established practice” or “company standard”, with no compendial or regulatory reference.
• No surveillance of monograph revisions, with specifications citing superseded editions.

Operational checklist

• Complete list of points of use, reconciled with the P&ID, master batch records and cleaning procedures.
• For each entry: role of the water, product served, sterile or non-sterile nature, route of administration, process stage.
• Minimum grade assigned with a precise reference to section 5.1 or 5.2 of the EMA guideline, or to an approved risk assessment.
• Explicit identification of the downstream step to which removal of the water contribution is attributed, or of its absence.
• Approved in-house microbiological and endotoxin specifications, distinct from compendial attributes and with a documented derivation.
• Documentary evidence linked to each row: qualification, monitoring, trend, cleaning validation.
• Matrix owner, review frequency and link to change control defined.
• Regulatory watch on monograph and guideline revisions, with a documented impact assessment.

If you work on these topics, The Pragmatic GMP collects technical analysis and regulatory updates on pharmaceutical facilities, written in the same register as this article.

Key takeaways

• Monographs define what a water is, not where it is to be used. The use → grade mapping sits in guideline EMA/CHMP/CVMP/QWP/496873/2018 in Europe and in no binding text in the United States, where it rests entirely with the user.
• Intended use is the primary criterion: sterility of the product, route of administration, process stage. The installed system is not a criterion.
• A compendial requirement and a GMP requirement for system control are different things: monograph compliance does not demonstrate control of the installation, and choosing WFI imports permanent system obligations.
• The 100 cfu/mL and 10 cfu/100 mL levels in USP <1231> are informational and non-binding: they must be translated into in-house specifications justified with the site's own data.
• The final rinse must be decided with an explicit criterion supported by evidence: none of the regulatory sources cited here sets its grade.
• The use → grade → rationale → evidence matrix is a controlled document with an owner, periodic review and change control, not a static annex.

Regulatory and technical references

• EMA/CHMP/CVMP/QWP/496873/2018 Guideline on the quality of water for pharmaceutical use, in force since 1 February 2021, sections 4.2, 5.1 and 5.2 — ema.europa.eu
• EudraLex Volume 4, Annex 1, in operation since 25 August 2023, clauses 2.5(v), 6.10 and 6.15 — health.ec.europa.eu
• Ph. Eur. 12th Ed., Issue 12.3 applicable from 1 July 2026: monographs 0008 Water, purified, 0169 Water for injections, 2249 Water for preparation of extracts; monograph 1927 Water, highly purified suppressed effective 1 April 2019; general chapters 2.2.38 and 2.2.44
• USP: Purified Water and Water for Injection monographs; chapters <643>, <645> and <1231> Water for Pharmaceutical Purposes (informational, official since 1 December 2021)
• 21 CFR 210.3(b)(3) definition of component; 21 CFR 211.48 Plumbing; 21 CFR 211.113 — ecfr.gov
• WHO TRS 1033, Annex 3 (2021) Good manufacturing practices: water for pharmaceutical use — who.int
• ICH Q9(R1) Quality Risk Management — ich.org

This article is part of the Pharmaceutical Water & WFI Systems cluster, covering design, generation, distribution, monitoring and qualification of pharmaceutical water systems.