Last updated: August 28, 2026
We may earn a commission when you buy through links on this page, at no extra cost to you. How we research and rate.
The short version
Over the last month we read the full product listing for 38 water filters across seven categories: reverse osmosis, countertop, whole house, under-sink carbon, well water, shower and refrigerator. Not the titles. The whole listing, every feature bullet.
We were answering one question each time: which NSF/ANSI standard does this product actually name, and does that standard cover the contaminant the buyer is worried about?
The answer varies enormously by category, and not in the direction price would predict. A $28 refrigerator cartridge names its certifier and two standards. A $493 whole house system names nothing at all.
What we counted
Seven categories, 38 products, one consistent test.
For each listing we recorded whether a certification claim appeared in the product title, whether it appeared anywhere in the body, whether a specific NSF/ANSI standard number was named, whether an accredited certifying body was named, and whether any tested figures were published.
Here is how the categories came out, best documented first.
Under-sink carbon filters. Five of six named a genuine performance standard, and two named the accredited body that issued it. The best of them stated the four attributes tested. This is the best-documented category in household water treatment, and it is not close.
Refrigerator filters. Four of five named specific standards, and three named all of 42, 53 and 401. The most detailed listing in the entire study is here, and it is an aftermarket cartridge rather than a manufacturer’s own.
Reverse osmosis. One of five named NSF/ANSI 58, the standard that actually tests RO performance. The most expensive system in that comparison carried 372 only, which covers lead-free materials.
Countertop filters. One certified to a standard, one tested against a standard by a named laboratory, one naming a laboratory without a standard, and one making no claim whatsoever at the highest price in its group.
Whole house systems. Zero of seven named a contaminant reduction standard. One named 372 for materials. The rest named nothing.
Well water systems. Zero named a standard number, though two published measured reduction figures in parts per million, which is arguably better.
Shower filters. Zero named a standard. One claimed 99 percent removal of hard water, which independent testing does not support.
Why certification tracks the product, not the price
The pattern above looks arbitrary until you consider what the test involves, and then it is almost entirely predictable.
NSF/ANSI performance certification means running a defined challenge water through the product at a defined flow rate, to the end of its rated life, in an accredited laboratory. The cost of that test scales with the flow rate and the rated capacity.
A refrigerator cartridge passes half a gallon a minute and is rated for 200 gallons. Testing it is cheap, and the result is a clean marketing asset. So almost everything in that category is certified, including the budget aftermarket options.
A whole house system passes ten gallons a minute and is rated for hundreds of thousands of gallons. The same test becomes expensive enough that nearly nobody on a marketplace bothers, and the contaminant profile varies so much by municipality that a single challenge water is a poor proxy anyway.
That is why a $28 cartridge can be better documented than a $493 system without either company behaving badly. It is also why the useful question changes by category: in refrigerator filters you should refuse anything without a standard number, and in whole house you would have nothing left to buy.
One consequence worth carrying with you. Certified whole house and well water systems do exist. They are mostly sold direct rather than on Amazon, because they are sized against your water test and often installed by the dealer. If a certificate is your requirement in those categories, the marketplace is the wrong channel.
Microplastics
The contaminant with the fastest-growing search interest and one of the clearest answers.
The standard is NSF/ANSI 401. Microplastic reduction sits within the emerging contaminants standard, so a filter certified to 401 has been tested for it. A filter certified only to 42 has not.
In our reading, the products naming 401 were concentrated almost entirely in refrigerator filters, where three separate products carried it. One explicitly claims 99 percent reduction of microplastics including particles far below the visible range, and describes itself as the first among leading brands certified for it.
Reverse osmosis handles microplastics as a matter of physics rather than certification. An RO membrane filters to around 0.0001 micron, which is orders of magnitude smaller than any microplastic particle. Nothing of that size passes. Several countertop and under-sink RO listings mention microplastics for this reason.
A carbon block at 0.5 micron will also physically retain most microplastic particles, since the particles of concern in drinking water are generally larger than that. The difference is that a 401 certificate means it was measured, and a micron rating means it should work.
Where to look: a 401-certified refrigerator filter if you drink from the door, an under-sink carbon block at 0.5 micron for the kitchen tap, or reverse osmosis if you want it dealt with by pore size rather than by adsorption.
Lead
The contaminant with the most misleading certification landscape, because of one number.
NSF/ANSI 53 covers lead reduction. NSF/ANSI 372 does not. Standard 372 certifies that a product’s own wetted parts are lead-free, meaning the housing is not adding lead to your water. It says nothing about removing lead that arrives in the supply.
Both are advertised as “NSF certified”, and in our reading 372 appeared far more often than 53 outside the under-sink and refrigerator categories. A $289 reverse osmosis system carried 372 with no performance standard at all.
Where lead comes from matters too. It rarely originates at the treatment plant; it leaches from service lines, solder and fixtures between the main and your tap. That has two implications. Your utility’s annual report will not tell you, because it is measured before your plumbing. And if your house predates the mid-1980s, a dedicated test is worth doing, because that is the one case where 53 rather than 42 genuinely matters.
Lead also arrives in two forms. Particulate lead is filtered mechanically by a fine carbon block. Dissolved lead needs adsorption or reverse osmosis. A filter certified to 53 for lead has been tested against both.
Where to look: an under-sink carbon filter certified to 53, or reverse osmosis, where one certified system publishes 98.9 percent lead reduction from its certificate.
Fluoride
The contaminant most often claimed and least often removed.
Carbon does not remove fluoride. It is a small dissolved ion, and activated carbon works by adsorbing larger organic molecules. This is settled and not controversial, and it did not stop several carbon products in our reading from listing fluoride among the things they reduce.
We found fluoride claimed on a countertop RO unit, a shower filter, an under-sink carbon system and a whole house filter. Of those, only the reverse osmosis unit is plausible.
Reverse osmosis does remove fluoride, and it is one of the few household treatments that does. One NSF/ANSI 58 certified system publishes a tested figure of 97.4 percent. Activated alumina is the other option, used in dedicated fluoride cartridges.
If fluoride is your reason for filtering, this is the single most important thing on this page: a carbon filter will not do it, whatever the listing says, and the presence of a fluoride claim on a carbon product is a reasonable signal to distrust the rest of that listing.
Where to look: under-sink reverse osmosis, or a countertop RO unit from our countertop guide if you cannot plumb one in.
PFAS, the forever chemicals
The contaminant where the certification framework is still catching up with the concern.
PFAS covers thousands of compounds, of which PFOA and PFOS are the two most commonly tested. There is no single NSF/ANSI standard dedicated to PFAS in the way 58 covers reverse osmosis; reduction claims appear under 53 and 401, and many manufacturers rely on separate third-party laboratory testing instead.
That produces the pattern we saw: PFAS claims were common and PFAS certifications were rare. The strongest documentation we found was a refrigerator cartridge publishing 95.3 percent reduction of PFOA and PFOS attributed to third-party testing, alongside its NSF/ANSI 401 certification for named pharmaceutical compounds.
On mechanism, both reverse osmosis and dense activated carbon reduce PFAS meaningfully, with RO generally more thorough. Granular carbon is considerably less effective than a carbon block.
Our practical reading: treat a bare “reduces PFAS” claim as unverified, and look for either a named percentage with a named laboratory, or certification to 53 or 401 with PFAS listed among the contaminants covered.
Where to look: reverse osmosis for the most thorough reduction, a 401-certified refrigerator filter or a 53-certified under-sink block otherwise.
Chlorine, chloramine and taste
The most common actual complaint, and the easiest to solve.
NSF/ANSI 42 covers this and nothing else. It is a genuine certification and it is the floor rather than an achievement, because nearly every carbon product carries it.
The distinction worth knowing is between chlorine and chloramine. Many US utilities have switched to chloramine, which is more stable in the distribution system and considerably harder for carbon and KDF to remove. A product certified to 42 for chlorine has not necessarily been tested for chloramine, and in our reading almost none of them distinguished between the two.
If your water tastes of pool and your utility uses chloramine, look for a product that names chloramine specifically. Catalytic carbon handles it better than standard activated carbon, and ascorbic acid neutralises it outright, which is why vitamin C cartridges appear in shower filters.
Where to look: almost anything carbon-based, including the cheapest options in our under-sink and refrigerator guides. This is the one contaminant where a budget certified filter is genuinely sufficient.
Iron, sulfur and hardness
Three problems that are mostly aesthetic, mostly on wells, and mostly outside the NSF/ANSI framework.
Iron and manganese cause the orange and black staining on baths and laundry. No product we read carried a performance certification for them. Two well water systems did something better in practice: they published measured input and output figures, iron from a maximum of 3.0 ppm down to 0.01 ppm. A range in parts per million is more useful than a percentage, because it tells you the concentration the system was designed to handle.
Hydrogen sulfide, the rotten egg smell, is handled at low levels by carbon and iron media and at higher levels by air injection. If the smell appears only on the hot tap, the water heater’s anode rod is the more likely cause than the supply.
Hardness is not filtration at all. Calcium and magnesium in solution require ion exchange, meaning a resin bed and a brine tank. No cartridge removes them, and independent testing of shower filters has consistently found hardness unchanged. This did not stop one shower filter in our set claiming 99 percent removal of hard water.
Where to look: well water systems for iron and sulfur, and a whole-house softener, which is a separate appliance, for hardness.
The six phrasings that carried the most weight
Across 38 listings, a handful of constructions did nearly all the work of making a weak claim look like a strong one. None of them is untrue. Each one is easy to read past.
“Certified” with no number. The most common by a distance. The best-selling refrigerator filter on Amazon says it is “certified to NSF/ANSI drinking water treatment standards” and never says which. Since 42, 53 and 401 answer completely different questions, this tells you a test happened without telling you what was tested.
“Tested against” rather than “certified to”. A countertop unit states it was tested by a named laboratory against the NSF/ANSI 58 standard. That is real evidence: a competent lab ran the standard’s method on a sample. It is not certification, which is an ongoing relationship with an accredited body, with surveillance and a public listing, and which can be withdrawn. One is a snapshot commissioned by the manufacturer. The other is a third party still standing behind the product.
Standards in the title that the body does not support. One reverse osmosis system puts four numbers in its product title, including the two that matter most, then substantiates exactly one in the description, and it is the materials standard. The other certification it names in the body is FCC, which governs radio emissions from the electronics.
“Surface” and other scope words. A disposable nursing pad in a different category taught us to watch for these, and water has its own: a “GOTS certified surface”, a “certified filter” inside an uncertified system, “NSF/ANSI 42 certified components and materials” rather than a certified system. Each is accurate. Each covers less than it appears to.
“Up to 99% of over 1,000 contaminants.” Unfalsifiable, because “1,000 contaminants” is not a category any standard measures and “up to” absorbs any result. We saw versions of this sentence in five of the seven categories. Set it beside the alternative from a certified system, which reads: TDS 97.9 percent, lead 98.9 percent, fluoride 97.4 percent, asbestos over 99.7 percent.
“99.999%.” Five nines is a figure from microbiology, where it describes log reduction of organisms. Applied to chemical contaminant reduction on a reverse osmosis system, it is not a number any performance standard produces.
What the best listings did differently
The reverse is worth recording too, because a handful of listings were genuinely excellent and they shared four habits.
They published the figures from the certificate. One RO system lists eight tested reductions by contaminant and percentage. A refrigerator cartridge names five pharmaceutical compounds by name, bisphenol A, phenytoin, estrone, ibuprofen and naproxen, with a single reduction figure covering them. That is what a 401 certificate actually contains, and almost nobody reproduces it.
They gave numbers in the units that decide the purchase. A well water system states iron reduction as a range: a maximum of 3.0 ppm in, 0.01 ppm out. Compare that with a percentage. If your test comes back at 6 ppm, the range tells you immediately that this is the wrong product and a percentage does not.
They named the certifying body and what it tested. The single best-documented listing in the study says it is certified by IAPMO against NSF/ANSI 42 and 53 “for Material Safety, Structure Integrity, Flow rate, and Filtration”. Standard, body, and the four attributes evaluated.
They stated a limit. This turned out to be the strongest single predictor of a well-documented listing overall. Three manufacturers volunteered that their system does not reduce TDS. One added, in capitals, that if TDS reduction is what you want you should consider its reverse osmosis range instead. One well water system says that above its stated iron ceiling you should contact them for a different solution. One shower filter notes that TDS testing is not supported by the product at all, heading off buyers who would otherwise test it wrongly and conclude it was broken.
A company willing to tell you what its product cannot do is usually the one being careful about what it says it can.
How to read a claim in ten seconds
The whole method, compressed.
Find the number, not the word. “NSF certified” tells you nothing. “NSF/ANSI 53” tells you it was tested for health contaminants. If a listing gives you letters without a number, it has told you something was certified without saying what.
Check the number matches the product. 58 belongs on reverse osmosis. 42 and 53 belong on carbon. 372 belongs on any plumbing part and proves nothing about filtration.
Read the body, not the title. Titles are written to win searches. In our reading, several products claimed a standard in the title that the description never mentioned again, and one listed four standard numbers in its title while substantiating only 372 in the body.
Prefer a named certifier. NSF International, IAPMO R&T and WQA are accredited bodies. A product naming one has an ongoing relationship that can be withdrawn, rather than a one-off test report.
Distrust a claim the technology cannot support. Fluoride on a carbon filter, hardness on a shower cartridge, “99.999%” on anything. One implausible claim is reason to discount the plausible ones beside it.
Value a stated limit. The listings that told us what their product does not do were consistently the better-documented ones overall. Three separate manufacturers in our set volunteered that their system does not reduce TDS, and one pointed buyers to a different product line for the job.
What this method cannot tell you
Worth being explicit, because the whole page rests on reading rather than testing.
We read listings. We do not run a laboratory, we have not tested any of these products, and we cannot verify that a certificate exists or that a published percentage is accurate.
A missing claim is not proof of a missing certificate. Listing copy is written product by product, often years apart, and we found several brands naming a standard on one product and not on another that almost certainly shares the same certification. The absence tells you what you have been given to check, not what is true.
A named standard is not proof of honesty either. It is a claim a third party issued and can withdraw, which is a stronger position than an unaccompanied assertion, and it is not a guarantee.
What the method does surface reliably is the gap between what a product claims and what it substantiates, and that gap turned out to be large, consistent, and almost entirely invisible from a search results page.
Start with your water, not with a filter
The last thing, and the one that saves the most money.
On municipal supply, your utility publishes an annual Consumer Confidence Report listing what was detected and at what level. It is free and it converts “what should I worry about” into a list. The one thing it will not show is lead from your own plumbing, since it is sampled before your service line.
On a private well, nobody tests anything unless you pay for it. The CDC recommends testing at least annually for bacteria and nitrates plus local concerns. The results decide the entire system, and buying before testing is how people end up owning two.
Once you have the list, the mapping is straightforward: chlorine and taste means carbon, lead and cysts means a 53-certified filter, fluoride or nitrates means reverse osmosis, bacteria means UV, iron and sulfur means well-specific media, and hardness means a softener rather than a filter at all.
Nothing else on this page matters as much as knowing which of those you actually have.
Questions about water filter certifications
What is the difference between NSF/ANSI 53 and 372?
Standard 53 certifies that the filter reduces health-related contaminants such as lead, cysts and VOCs. Standard 372 certifies that the product’s own wetted parts are lead-free, so the housing is not adding lead. Only 53 is a filtration claim, and both are advertised as “NSF certified”.
Which filter removes microplastics?
Anything certified to NSF/ANSI 401 has been tested for it, and reverse osmosis removes them by pore size regardless of certification. A 0.5 micron carbon block will physically retain most of them too.
Do carbon filters remove fluoride?
No. Fluoride is a small dissolved ion that carbon does not adsorb. Reverse osmosis or activated alumina are the options. A fluoride claim on a carbon product is a reason to distrust the rest of that listing.
Is there an NSF standard for PFAS?
Not a dedicated one. PFAS reduction appears under standards 53 and 401, and many manufacturers rely on separate third-party laboratory testing. Look for a named percentage with a named lab, or 53 or 401 certification listing PFAS among the contaminants.
Why do cheap filters have better certification than expensive systems?
Because the cost of the test scales with flow rate and rated capacity. Certifying a 200 gallon refrigerator cartridge is cheap; certifying a whole house system at ten gallons a minute over hundreds of thousands of gallons is not.
Are WQA and IAPMO as good as NSF?
Yes, where they certify against the same standard. All three are accredited bodies testing to NSF/ANSI standards. The standard number matters, not which body issued the certificate.
Do shower filters soften water?
No. Hardness requires ion exchange with a resin bed and brine tank. Independent testing of shower filters has found hardness unchanged, and one product in our set still claims 99 percent removal.
Should I buy a filter certified to more standards?
Only if the extra standards cover something you have. A filter certified to 42, 53 and 401 is genuinely broader than one certified to 42 alone, and if your only complaint is chlorine taste then 42 is what you need and the rest is money spent on a problem you do not have.
Organic Connection is reader supported. We check every product against its stated certification and say plainly when a claim cannot be verified. How we research and rate.