Every roster page ends the same way. This is the inverse.

Each claim-based page on this site answers “which products hold this claim”. A question none of them asks is whether the claim exists at all.

So: EPA’s regulated contaminant tables, against NSF’s complete list of selectable reduction claims, both read on 2026-09-10.

Entries
EPA regulated contaminant entries 94
With an NSF reduction claim of that name 61
With no NSF claim of that name 33

Where the gaps are

Group Gaps
Organic chemicals 19
Microorganisms 4
Inorganic chemicals 4
Disinfection byproducts 3
Radionuclides 3

Three of those deserve attention because they are contaminants people are actively shopping for.

HAA5 — an enforceable limit with nothing to shop for

The five haloacetic acids are one of two regulated byproduct groups, capped at 0.060 mg/L, and they appear on every annual water quality report alongside trihalomethanes.

Trihalomethanes has a claim, held by 153 models. Haloacetic acids has none.

Filters certified for trihalomethanes

Three NSF claims sound close and are not the same group. Tribromoacetic acid is not one of the five EPA regulates — those are monochloroacetic, dichloroacetic, trichloroacetic, bromoacetic and dibromoacetic acid. Haloacetonitriles and haloketones are separate compound families.

So a household reading both byproduct figures off its report can shop against one and not the other.

Disinfection byproducts: both figures, and what they mean

PFAS — one compound is genuinely uncovered, and it is not the one you expect

EPA’s regulation covers PFOA, PFOS, PFHxS, PFNA and HFPO-DA individually, plus a Hazard Index combining four of them.

NSF offers named claims for PFOA and PFOS only — so on claim names alone, four EPA entries look uncovered.

They are not. NSF defines its Total PFAS claim in the listing footnotes, and the definition names the compounds:

Total PFAS includes perfluoroheptanoic (PFHpA), perfluorohexane sulfonic acid (PFHxS), perfluorononanoic acid (PFNA), perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), perfluorobutanesulfonic acid (PFBS), perfluorodecanoic acid (PFDA)

PFHxS, PFNA and PFBS are in there. So the real gap is a single compound: HFPO-DA, the one EPA calls GenX, which appears in no NSF claim by name and is not among the seven in the bundle.

Two caveats hold. A product must actually hold the Total PFAS claim for the bundle to apply, and 49 models did. And EPA has stated that current certification standards do not yet indicate a filter will reduce PFAS to its limits, so appearing in a definition is not evidence of meeting the regulation.

Filters certified for PFAS, and the three separate claims

Organic chemicals — 19 of the 33

The biggest of the five groups, and mostly pesticides and industrial compounds: acrylamide, benzo(a)pyrene, dalapon, dichloromethane, the two di(2-ethylhexyl) compounds, dioxin, diquat, endothall, epichlorohydrin, glyphosate, oxamyl, picloram and vinyl chloride among them.

Several have a partial route. NSF’s VOC reduction claim is a bundle covering a defined list of compounds, so a product holding it may address some of these without any of them being named individually.

What a VOC claim covers

What the absence means, and what it does not

This is the part that matters most, and it cuts against the alarming reading.

A missing claim is a missing verification, not a missing technology. EPA maps contaminants to treatment types — distillation, reverse osmosis, ion exchange, adsorptive media — independently of whether NSF offers a claim. Reverse osmosis reaches a great deal of this list.

Distillation is invisible here entirely. NSF offers no selectable reduction claims under NSF/ANSI 62, its distillation standard, so no distiller appears in any claim search — not on this page and not on any roster on this site.

Microorganisms have a separate route. NSF/ANSI 55 certifies ultraviolet systems for Disinfection Performance Class A or B rather than naming organisms, which is why coliforms, viruses, Legionella and heterotrophic plate count show as gaps while a certification path exists.

So read the 33 as “no product can be certified and listed for this by name”, which is a real constraint on a buyer and not a verdict on treatment.

The decision framework

1. Check whether your contaminant has a claim before shopping

If it does, the listing settles it per model.

Which claims exist at all, and how populated each is

If it does not, no amount of product comparison will produce a certified answer, and that is worth knowing in the first five minutes rather than after a purchase.

2. Fall back to the technology mapping

EPA names treatment types per contaminant. That is weaker than a certified claim — it describes a category, not a tested model — and it is what remains when no claim exists.

3. Ask the supplier the two questions

Which accredited body certified this, and for which claim. If the answer names a contaminant with no claim, the assertion is not something any listing can support.

How to read an NSF listing, footnotes included

4. Treat reverse osmosis as the broad instrument

Under NSF/ANSI 58 only TDS reduction is required and everything else is optional, so an RO system is not automatically certified for anything in particular — but it is the technology EPA maps to the widest range of this list.

The decision path

Your contaminant Claim exists? What to do
Lead, cyst, TTHM, nitrate, fluoride Yes Search the claim, check the model
Haloacetic acids No No certified option; utility-side control
PFOA or PFOS Yes, by name 60 models each
PFHxS, PFNA, GenX No claim of that name Total PFAS claim is the nearest
A regulated pesticide Usually no Check whether VOC bundle covers it
Uranium or alpha/beta particles No Technology mapping, not certification
Coliforms, viruses, Legionella No reduction claim NSF/ANSI 55 disinfection classes

Choose a certified product if

  • Your contaminant has a claim. That is 61 of 94 EPA entries, including most of what households actually test for.

Choose on technology mapping if

  • Your contaminant is among the 33. EPA’s treatment-type mapping is the next reference, and reverse osmosis or distillation will often be the answer even though neither can be certified for it by name.

Choose to ask rather than buy if

  • A product claims one of the 33 by name. No listing can support that, and the question of which body certified it is the right one before paying.

Mistakes that cost the most money

Assuming a claim exists because the contaminant is regulated. A third of EPA’s regulated entries have none.

Reading a missing claim as “nothing removes it”. It means nothing is certified for it by name, which is a different statement.

Accepting a marketing claim for one of the 33. There is no listing to check it against, which is precisely why it can be printed.

Assuming a TTHM claim covers haloacetic acids. Different group, no claim, and the similar-sounding NSF claims are other compound families.

Assuming a total PFAS claim covers the four unnamed PFAS. NSF does not enumerate what it covers, and EPA has said certification does not yet demonstrate reduction to its limits.

Ruling out distillation because it never appears in certified lists. It cannot appear; there are no claims under its standard.

Three situations, and what changes

Your report shows HAA5 above TTHM. The shopping question has no certified answer for HAA5. What remains is the utility-side control — the Operational Evaluation Level and the report that follows it — rather than a product.

You are buying for PFAS after the 2024 rule. PFOA and PFOS are named and searchable. The other four entries are not, so a product addressing them specifically cannot be verified in NSF’s listing today.

A salesperson names a contaminant from the list of 33. Ask which body and which claim. The honest answer will be that no such claim exists, and what happens next tells you what kind of supplier you are dealing with.

The objections worth answering

“This makes certification look useless.” The opposite: 61 of 94 entries have a claim, including nearly everything a household tests for. Knowing which third is uncovered is what makes the other two-thirds worth relying on.

“Reverse osmosis removes almost all of these anyway.” Quite possibly, and EPA’s technology mapping supports that for many. The point is that you cannot verify it per model, which is the whole function of a listing.

“You have only checked NSF.” Correct, and stated. WQA and IAPMO certify to the same NSF/ANSI standards and publish separate listings, and their claim taxonomies were not compared here.

“These figures will move.” They will. NSF dates its listings and asks readers to confirm status at source, and EPA revises its regulations — the 2024 PFAS rule is why four of these gaps exist at all.

Who this is for

Anyone who has looked for a certified filter for a specific contaminant and found nothing, and anyone evaluating a product claim that no listing seems to support.

It names no recommended product and is a map of where verification exists.

What to settle first

Does a claim for my contaminant exist at all?

Sixty-one times out of 94 the answer is yes and the listing takes over. The other 33 times, knowing it early saves the search.

Which claims exist at all, and how populated each isHow to read an NSF listing, footnotes includedCity water treatment, and the three gaps a utility cannot closeThe VOC bundle: broad, printed in full, and it does not reach the gapsArsenic: a claim exists, and only for one of the two speciesManganese: health guidance, no federal limit, and no claim to searchRadium and uranium: three of four regulated radionuclides have no claimEPA names a technology for many of these; certification is a separate questionUnregulated contaminants: the candidate list and the monitoring ruleMicrocystin: unregulated, and certified anywayRadon: a certified claim for a contaminant with no limitLegionella: regulated, unmeasured, and uncertifiedDistillation, which reaches some of these and certifies against noneThe inverse case: unregulated contaminants with large certified fieldsColiform and E. coli: regulated, measured, and uncertifiable