Four NSF/ANSI numbers, four different questions — and what an unmarked unit has not proven
NSF/ANSI 42, 53, 55 and 58 are four separate standards, and each answers a different question about a drinking-water treatment unit: whether it changes taste, odour and appearance (42), whether it reduces named health-related contaminants (53), whether it is an ultraviolet microbiological treatment system and of which class (55), and whether it is a reverse-osmosis system (58). They are not interchangeable, and a unit carrying one has not, by that mark, been assessed to another. What this site holds, and what it does not, is stated once on the Certifications page rather than repeated here. This article is about how to read the four standards, not about what to claim from them.
Why four numbers get treated as one thing
The four standards are written and published separately, and they are numbered separately, but they arrive at a buyer in the same shape: a short string that starts with NSF/ANSI and ends in a two-digit number. Sellers list them in the same row of the same table, as though the row were a single grade that a product either has or lacks. It is not. Each standard was written for a different kind of product or a different kind of question, and the number is the only part that looks alike.
There is a second reason the four blur together. A UV unit sold into a drinking-water line sits next to reverse-osmosis systems, carbon filters and softeners, and the same marketing language gets used across all of them — certified, tested, approved — without saying certified to what, tested for what, or approved by whom. The standard number is the part that makes a claim checkable. Take it away and what is left is an assurance, not a specification.
So the useful move is not to memorise which number is “better”. It is to keep four questions apart, because the standards do:
42 — what the water looks, tastes and smells like
Standard 42 covers aesthetic effects: the things a person notices with their senses. The claims made under it concern taste and odour, and particulate matter — the material that makes water look cloudy or leaves a residue. It is a real and useful standard. It is also the one most often mistaken for something it is not.
The distinction that matters is in the name. An aesthetic effect is a matter of how water is experienced; it is not a statement about what is dissolved in it or what is living in it. A unit that carries a 42 claim has been evaluated for the sensory claim it makes, and that evaluation says nothing about health-related contaminants and nothing about microorganisms. Water can taste and look entirely clean while carrying something a 42 claim was never written to address — which is why the article on why appearance is a poor guide to what is in water is a companion to this one rather than a repeat of it.
When a page says “NSF/ANSI 42 certified” with no further detail, the useful next question is which claim within the standard. The standard is organised as a set of claims, not as a single pass-or-fail; a unit is certified for the claims it was tested against, and those are the only ones it can be described with.
53 — named health-related contaminants
Standard 53 covers health effects. Where 42 is about the senses, 53 is about substances that are harmful to health, and it is written the same way in both cases: as a list of specific, named claims. Lead. Volatile organic compounds. Certain pesticides. Certain disinfection by-products. The organisms Cryptosporidium and Giardia. Arsenic. Each is a separate claim with its own test, its own influent challenge water and its own reduction requirement.
What follows from that structure is the single most useful thing to understand about the whole family of standards: a certification is a statement about a named claim under stated conditions, not a grade attached to a product. A unit certified to reduce lead is not, by virtue of that, certified to reduce arsenic. A unit certified to reduce three contaminants out of the list is described by those three — the rest of the list is simply absent from its certification, in both directions.
So the two-part question to ask about any claim under 53 is which contaminant, and under which influent water. The second half is where most of the substance lies. A reduction percentage is meaningless unless the challenge water it was measured against is named, because the result belongs to that water and to those operating conditions. This is the same principle that makes a dose figure without its test conditions unreadable, and it applies across the family.
55 — ultraviolet microbiological treatment, and its two classes
Standard 55 is the one of the four written specifically for ultraviolet microbiological water treatment systems, and it is the standard most often invoked when UV units are compared. Its defining feature is that it does not treat all UV systems as one thing. It separates them into two classes, and the separation is not a quality ranking — it is a statement about the water the unit is intended to receive.
Class A is for systems intended to disinfect water that may be microbiologically contaminated. A Class A system is tested against a microbiological challenge, and the standard sets an input-water condition it must be capable of handling. Class B is for systems intended to provide supplemental bactericidal treatment of water that is already considered safe — a different job, tested a different way, with no pretence of disinfecting a contaminated supply.
Two consequences follow. The first is that a class is only meaningful with its input condition attached: a statement that a unit is “Class A” describes the tests it was put through, and those tests were defined against a stated incoming water. A Class A system installed on a supply outside that condition is not operating under the conditions of its own class. The second is that Class A is not a claim that no upstream treatment is needed. It is part of a treatment train, and it assumes the train it was written for.
The classification itself — what A and B mean — is stated once on this site in the FAQ, in a paragraph, because that is a question with a short answer. What belongs here is the part the paragraph cannot carry: what the class commits the unit to, and what it leaves to everything upstream of it.
58 — complete reverse-osmosis systems
Standard 58 covers reverse-osmosis drinking-water treatment systems, and the word doing the work is systems. It is written for a complete assembly — the membrane, the pre-treatment it depends on, any storage, and any post-treatment stage that ships with it — rather than for a single component inside one. A claim under 58 is a claim about an assembled product as it is supplied and installed.
That has a practical consequence for anyone reading a specification for a component. A part of a reverse-osmosis system — a membrane, a post-treatment stage, a UV unit on the permeate line — is not, by itself, a 58 system, and a claim against 58 on the assembly does not transfer to the parts it was built from. Conversely, a part supplied to a system builder is assessed under whatever standard is written for that part, not under the system standard. The question “is this component 58 certified” is therefore malformed: 58 is a standard for the finished system.
This is where the four standards stop looking interchangeable at all. Two of them (42, 53) attach claims to a treatment unit and its declared reductions; one (55) attaches a class to a UV system and the water it expects; one (58) attaches a claim to an assembled RO product. Nothing in that set substitutes for anything else in it.
Why a mark to one says nothing about another
Four differences explain why the standards cannot be read across from one to the next.
Different objects. 42, 53 and 55 attach claims to a treatment unit of a particular kind; 58 attaches its claim to a whole system. A standard written for one object does not describe another.
Different claim structures. 42 and 53 are built from itemised claims — a unit holds the ones it was tested for. 55 is built from classes. 58 is built around a complete product. Itemisation means partial coverage is normal and expected, and a partial certificate is not a deficiency; it is the ordinary form.
Different test conditions. Each standard defines its own challenge water, flow, temperature and acceptance basis. A result obtained under one standard's conditions is not a result under another's, even where the measured quantity has the same name — a transmittance figure being the clearest example of a number that belongs to its test water rather than to the product.
Different issuing bodies. A mark is issued by a certification body against a standard, and the body that issued it is part of the claim. Two certificates can cite the same standard number and differ in who signed them, in which edition of the standard they were assessed against, and in whether the assessment was per model or per family. Those details are not paperwork noise; they are the content of the claim.
What an unmarked unit has, and has not, proven
“Not marked” and “failed” are not the same statement, and a reading guide that discusses the standards at all has to hold both sides of the question at once. There are several ordinary reasons a product carries no mark to one of these standards, and none of them is automatically a defect:
It was never submitted. Third-party certification is a programme a manufacturer chooses to enter, market by market, at a cost per model and per claim. A unit built for markets where the mark is not the required path may never have been put through it.
The mark is not the regulatory path in that market. In the United States, for instance, the federal entry point for a UV device can be an establishment registration rather than a product certification — a distinction set out in a separate article, because it is routinely misread. A manufacturer can be correctly registered and hold no product certification; those two facts coexist without contradiction.
The standard does not describe the product. A component supplied into someone else's system is not the object that 58, for example, was written for. Its absence of a mark is a statement about the standard's scope, not about the component.
Internal testing happened and third-party certification did not. A manufacturer may hold its own measurements without having them verified externally. That is not nothing — but it is also not the same thing, and it should never be described using the certification's vocabulary.
What an unmarked unit has not proven is the thing the standard exists to prove: that an independent body assessed a named model against a named claim under named conditions. That gap does not make the product bad. It makes the product unverified against that standard, which is a different and narrower statement — and the honest form of it is the one worth insisting on.
Four questions that settle it from the paperwork
The point of a standard number is that it makes a claim checkable. Four questions turn a certificate from a document into a verifiable statement, and all four can be answered from paperwork alone — no laboratory work, no site visit.
One — which standard, and which edition? A standard number without its edition is incomplete, because the requirements change between editions. Ask for the number and the year, and check that the two match on the certificate.
Two — which model, and which claim within the standard? Ask for the model designation as it appears on the certificate, and for the specific items claimed — the named contaminants, the class, or the system configuration. A certificate that names a family while the specification names a model is the commonest mismatch, and it is usually innocent, but it has to be resolved.
Three — under which conditions? For a reduction claim, this is the influent challenge water and the operating conditions. For a class, it is the input-water condition the class assumes. A number without its conditions is not a weak answer; it is not an answer.
Four — issued by whom, and may I see it? The certification body is part of the claim, and so is the scope statement on the certificate. A document that can be produced and read is the whole point; where it cannot be produced, the claim is an assurance and should be treated as one. The companion article on reading a supplier's certificate works through the same four questions from the certificate side.
Questions this article raises
Does a UV unit need all four standards?
No. They describe different products and different questions. A UV stage on a point-of-use line is the object of 55, not of 58, and 42 and 53 are relevant only where the unit makes an aesthetic or a health-related claim. Building a checklist of all four and requiring every box is a common procurement habit and it does not correspond to how the standards are written.
If a unit is not certified to 55, is it unproven as a UV unit?
It is unverified against that standard, which is a narrower statement than “unproven”. 55 is the certification path for UV microbiological treatment; a manufacturer that has not entered it holds no certification under it, and should not describe the product in its vocabulary. Whether that matters depends on your programme — where a mark is a hard requirement, it belongs in the specification from the outset, because nothing other than a certified unit can answer it.
Class A or Class B — which should I ask for?
Ask first what water the unit will receive. Class A is written for water that may be microbiologically contaminated; Class B is written for supplemental treatment of water already considered safe. The class should follow the supply, not the price list, and if the supply falls outside the input condition a class assumes, the class is not doing the work it appears to do.
Is the manufacturer's own test report the same as a certification?
No, and the difference is the whole value of the exercise. A manufacturer's own report is that manufacturer's measurement; a certification is an independent body's assessment against a published standard. Both can be useful, and they are not interchangeable — a self-report should never be described with the certification's language, and a specification written as though the two were equivalent cannot be checked by anyone.
Our programme requires one of these marks. When should that be raised?
At the outset, and in writing. A mark is a hard requirement for some programmes and a preference for others, and the two lead to different shortlists — establishing which one applies before the specification is written saves the work of discovering it afterwards. Ask for the certificate and its scope statement as part of the first exchange, and treat an assurance that a mark is held, offered without the document, as an unverified statement rather than as a fact.
Ask for the document, not the adjective
If you are comparing units and a standard number appears in a specification you have been given, we will tell you plainly what we hold and what we do not — the same answer that appears on the Certifications page. If a mark to one of these standards is a requirement of your programme, say so in the first message and we will confirm the position rather than let the process run.
- Models: ZXSJ-UVC-1GPM (1 GPM / 4 LPM, 1/4") ｜ ZXSJ-UVC-3GPM (3 GPM / 11.4 LPM, 3/8")
- Available now: instruction manual · parameter table · CE certificate DPCW/01068/06/KA/26 · EPA Establishment No. 105777-CHN-1
- On request: a written statement of what is and is not held, including the standards deliberately not claimed — the same list is kept in one place on the Certifications page
Send an enquiry
Typical response within one business day.
This form is not yet connected to a mail service — the enquiry endpoint will be wired up when it is connected. Until then, enquiries reach the manufacturer through the contact page.
Sources and scope
4 sources
- The descriptions of NSF/ANSI 42, 53, 55 and 58 given here are summaries of what each standard is scoped to, in the terms the standards themselves use — aesthetic effects, health effects, ultraviolet microbiological water treatment systems and reverse-osmosis drinking-water treatment systems. The standards are the authority for their own requirements; nothing on this page should be read as a restatement of a clause, a limit or an acceptance criterion from any of them.
- The two classes of NSF/ANSI 55, and the input-water condition each class assumes, are described at the level of the framework only. The scope statement on a certificate governs the claims that certificate supports, and the edition of the standard cited on it governs the requirements that were applied.
- No certification under any of these standards is held by SKWEIT® for any model, and no mark is displayed. What is held, and the statements deliberately not made, are listed in one place on the Certifications page.
- This article is a reading guide for buyers, not legal, regulatory or procurement advice, and it draws no comparison between suppliers. Where a standard number appears in a specification you are assessing, the standard text and the certificate's own scope statement govern; a summary of the kind given here does not.