Every arsenic claim comes with a condition
Most certifications are unconditional: the model holds the claim, and that is that.
Arsenic is not. NSF attaches a footnote to these claims:
* This product is intended for use only on water supplies that have been shown to be at or below the referenced concentration and valency of arsenic.
Two conditions, and both fall on the buyer. You must have shown that your arsenic is below the ceiling in the claim, and that it is the species the claim covers.
Neither is knowable without a laboratory test that reports both.
→ Arsenic: the health effects, prevalence and testing
The species condition is absolute
Every arsenic claim NSF offers is for pentavalent arsenic. There is no trivalent claim under any standard.
That matters because the two behave very differently in treatment. New Hampshire DES states that without pre-oxidation, reverse osmosis removes “typically 60 to 65 percent” of arsenic III, against “greater than 98 percent” of arsenic V.
So a certified system facing the wrong species is not a certified system facing a harder job — it is outside the terms of its own certification.
Where trivalent arsenic is at or above 10 ppb in raw water, NHDES advises adding a pre-oxidation step, and lists chlorine, ozone, permanganate and solid manganese dioxide media as effective for that conversion. Air, UV light and chlorine dioxide it lists as not effective.
Three claims, and they barely overlap
| Claim | Standard | Distinct models | Companies |
|---|---|---|---|
| Pentavalent, to 50 ppb | 53 | 15 | 6 |
| Pentavalent, to 50 ppb | 58 | 11 | 2 |
| Pentavalent, to 300 ppb | 58 | 110 | 33 |
135 distinct models in total. The surprise is that they are not nested.
Of the 11 models holding the NSF/ANSI 58 claim to 50 ppb, one also holds the 300 ppb claim. Of the 15 holding the NSF/ANSI 53 claim to 50 ppb, none holds the 300 ppb claim.
That is unlike the other NSF/ANSI 58 claims, where nitrate sits inside hexavalent chromium, which sits inside fluoride and lead. Here a claim at one ceiling tells you nothing about the other.
→ The nested NSF/ANSI 58 claims, for contrast
So match the claim to your measured figure. A system certified to 300 ppb was tested against that influent concentration; one certified to 50 ppb was not.
Where the certified products are
Almost entirely under-sink reverse osmosis on a separate tap: 110 of the 110 models holding the 300 ppb claim, and 11 of 11 holding the NSF/ANSI 58 claim to 50 ppb.
The NSF/ANSI 53 set is the exception, and a small one: 13 plumbed-in to separate tap, one counter-top and one pour-through. Those two are the only certified arsenic options outside a fitted system in this listing.
The decision framework
1. Test for arsenic and for species
Both, in one report. Without the species, none of these certifications applies to your water by its own terms.
Minnesota Department of Health advises testing a well for arsenic at least once beyond the test performed when the well was constructed, and reports about 10 percent of Minnesota private wells above 10 µg/L.
2. Read your figure against the claim ceilings
EPA’s enforceable limit is 0.010 mg/L — 10 ppb.
If your figure is under 50 ppb, both ceilings are candidates and the 50 ppb claims are matched to your range.
If it is between 50 and 300 ppb, only the 300 ppb claim was tested against that influent, and 110 models hold it.
If it is above 300 ppb, no claim here covers your influent concentration. That is a professional treatment question rather than a shopping one.
3. Deal with species before choosing a product
If the arsenic is pentavalent, the claims apply as written.
If trivalent is present at or above 10 ppb, NHDES advises pre-oxidation first. The certified filter goes downstream of that step, not instead of it.
4. Confirm the claim on the model
Under NSF/ANSI 58, TDS reduction is the only required claim; arsenic is optional and listed per model. NSF also lists arsenic reduction and pentavalent arsenic reduction as separate claim names.
→ How to read an NSF listing, footnotes included
Applying the framework: who holds which claim
To 300 ppb under NSF/ANSI 58 — 110 models, 33 companies. The broad field, and the same dealer-channel names that hold the nitrate and fluoride claims.
To 50 ppb under NSF/ANSI 58 — 11 models, 2 companies. A narrow field, and only one of the 11 also carries the 300 ppb claim.
To 50 ppb under NSF/ANSI 53 — 15 models, 6 companies. The only set containing anything that is not plumbed in: one counter-top and one pour-through.
Notes tied back to the criteria
- The 300 ppb claim is the populated one, which is counterintuitive until you read it as the harder test: it was validated against a higher influent concentration.
- Non-overlap is the practical point. Holding a 300 ppb claim and holding a 50 ppb claim are separate certifications, and 134 of the 135 models hold only one of them.
- Two portable options exist. They sit in the NSF/ANSI 53 set and nowhere else.
- Nothing here separates models on performance. NSF records claim membership, not a removal percentage.
The decision path
| Your situation | What decides it | Where that lands you |
|---|---|---|
| No species result | The valency condition | Test again; no claim applies yet |
| Pentavalent, under 50 ppb | Matched ceiling | 26 models across two standards |
| Pentavalent, 50 to 300 ppb | Only one ceiling fits | 110 models under NSF/ANSI 58 |
| Pentavalent, above 300 ppb | Outside every claim | Professional treatment |
| Trivalent at or above 10 ppb | Pre-oxidation first | Then a certified filter downstream |
| Cannot fit an under-sink unit | Two non-plumbed options | Both in the NSF/ANSI 53 set |
| Well never tested for arsenic | Nothing to size | MDH advises testing at least once more |
Choose a 300 ppb claim if
- Your measured pentavalent arsenic sits above 50 ppb, where it is the only ceiling tested against your influent.
- You want the widest choice, at 110 models from 33 companies.
Choose a 50 ppb claim if
- Your figure sits comfortably below 50 ppb and you want a claim matched to your range.
- You need a counter-top or pour-through option, which exist only here.
Choose pre-oxidation first if
- Trivalent arsenic is at or above 10 ppb. NHDES advises the step, and the certification downstream assumes it has happened.
Choose to test before anything if
- You have a number but no species, or no number at all. Every claim on this page is conditional on both.
Mistakes that cost the most money
Buying without a species result. The most expensive mistake here, because the certification explicitly excludes supplies whose valency has not been shown.
Assuming a higher ceiling covers a lower one. 134 of 135 models hold one claim, not both.
Treating reverse osmosis as automatically arsenic-certified. Only TDS reduction is required under NSF/ANSI 58; arsenic is optional.
Skipping pre-oxidation with trivalent arsenic present. NHDES puts untreated RO removal of arsenic III at typically 60 to 65 percent.
Reading the ceiling as a treated-water figure. It is the influent concentration the product was tested against.
Relying on a neighbour’s result. MDH states arsenic levels can vary between wells even within a small area.
Three situations, and what changes
A well test showing arsenic at 30 ppb, species not reported. The common case, and not yet actionable. Ask the laboratory for speciation before choosing anything; the answer decides whether pre-oxidation is part of the job.
Arsenic at 120 ppb, confirmed pentavalent. Only the 300 ppb claim was tested against that influent. 110 models hold it, and the decision becomes ordinary.
A rental, or no space under the sink. Two certified options exist — one counter-top and one pour-through, both in the NSF/ANSI 53 set to 50 ppb. Check your figure is inside that ceiling.
The objections worth answering
“Why is the 300 ppb claim more common than the 50 ppb one?” Read as a test rather than a limit, 300 ppb is the harder validation. More manufacturers have sought it. NSF records what was certified, not why.
“My filter says it removes 99 percent of arsenic.” That names a figure and no species, no influent concentration and no standard. All three are in the listing.
“Species testing costs extra.” It does, and every certification on this page is void without it by NSF’s own wording. It is the cheaper half of the decision.
“These figures will move.” They will. NSF dates its listings and asks readers to confirm status at source, which is why the capture date sits beside every figure here.
Who this is for
Anyone with arsenic on a water test choosing a filter, and anyone who has been quoted a system without being asked which species they have.
It is not the contaminant page, and it names no recommended product.
What to settle first
What is my arsenic figure, and is it trivalent or pentavalent?
Both answers are required before any certification on this page applies. Without the second one, no product here is certified for your water by its own terms.
→ Arsenic: health effects, prevalence and testing → How to read an NSF listing, footnotes included → Which claims exist at all, and how populated each is → The valency condition, as EPA writes it into the regulation