Only two NSF certifications actually prove a drinking-water filter reduces PFOA and PFOS to a measured limit: NSF/ANSI 53 with the P473 protocol for carbon-block media, and NSF/ANSI 58 for reverse-osmosis systems with the PFOA/PFOS reduction claim. Everything else on the box, NSF 42, "tested to," "filters 99% of contaminants", is marketing.
NSF International (now operating as "NSF") runs the dominant third-party certification scheme for North American drinking-water treatment products. Its standards are written jointly with ANSI (the American National Standards Institute) and are referenced directly by EPA guidance, state plumbing codes, and almost every utility procurement document in the country. The problem for consumers is that there are five different NSF standards that can legitimately appear on a filter box, and only two of them say anything specific about PFAS.
Here is the short, accurate version. NSF/ANSI 42 covers aesthetic effects, chlorine taste, odour, and particulate reduction. It is the cheapest certification a manufacturer can claim and it says nothing about health-effects contaminants. A pitcher filter that lists only "Tested and Certified to NSF/ANSI 42" has not been tested for PFAS at all. NSF/ANSI 53 covers health-effects contaminants, lead, cysts, VOCs, chromium, and (when the P473 protocol is invoked) PFOA and PFOS. NSF/ANSI 58 is the standalone standard for residential reverse-osmosis systems and now includes its own PFOA/PFOS reduction claim. NSF/ANSI 401 covers fifteen "emerging compounds and incidental contaminants", mostly pharmaceuticals, an herbicide, BPA, and a flame retardant. NSF/ANSI 244 covers microbiological reduction (bacteria, viruses, cysts) for point-of-use units. Of those five, only 53 P473 and 58 carry an explicit PFAS reduction claim.
| Standard | What it actually proves | Covers PFAS? |
|---|---|---|
| NSF/ANSI 42 | Aesthetic, chlorine taste, odour, particulates | No |
| NSF/ANSI 53 | Health-effects contaminants (lead, cysts, VOCs, Cr-VI) | Only if P473 claim is listed |
| NSF/ANSI 53 P473 | Protocol within 53 for PFOA + PFOS reduction | Yes, long-chain |
| NSF/ANSI 58 | Reverse-osmosis systems; PFOA/PFOS claim added | Yes, broad-spectrum |
| NSF/ANSI 401 | 15 emerging contaminants, pharmaceuticals, BPA, herbicides | No (PFAS not listed) |
| NSF/ANSI 244 | Microbiological reduction, bacteria, viruses, cysts | No |
If you take one thing from this guide, take this: a manufacturer can advertise "NSF certified" while only carrying a NSF/ANSI 42 listing for chlorine taste reduction. That certification is technically true and tells you absolutely nothing about whether the filter touches the six PFAS regulated under the April 2024 EPA MCLs.
P473 began life as a free-standing NSF "Protocol", the "P" prefix denotes a protocol developed by NSF outside the consensus-standard process. It was published in 2016 to give manufacturers a testable claim against the EPA Lifetime Health Advisory that had just been set at 70 ng/L combined PFOA + PFOS. The protocol was subsequently folded into the relevant consensus standards: a filter today can be certified to NSF/ANSI 53 with the P473 reduction claim for activated-carbon and similar adsorption media, or to NSF/ANSI 58 with the PFOA/PFOS reduction claim for RO systems.
The test protocol itself is specific. Filters are challenged with an influent concentration of 1.5 µg/L (1,500 ng/L) combined PFOA + PFOS, roughly 375 times the new EPA MCL, and must reduce that influent to a maximum effluent concentration of 70 ng/L or less throughout the manufacturer's rated cartridge life. Analytical confirmation is done by EPA Method 537.1 (or 533 for the short-chain compounds), which is the same isotope-dilution LC-MS/MS method used by every accredited PFAS lab in the country.
Two facts about this protocol matter for the post-April-2024 world. First, it is a real, defensible test: an actual challenge concentration on an actual flow rig with actual LC-MS confirmation, not a self-reported claim. Second, the 70 ng/L pass threshold is no longer aligned with federal drinking-water law. The EPA Maximum Contaminant Level finalised at 40 CFR §141.61(c) is 4.0 ng/L for PFOA and 4.0 ng/L for PFOS, seventeen times lower than the P473 pass criterion. A filter can clear P473 and still leave water above the MCL. In practice, well-designed carbon blocks reduce PFOA and PFOS to far below 70 ng/L (often non-detect at the start of cartridge life), but P473 alone does not guarantee MCL-level performance.
NSF/ANSI 58 governs residential reverse-osmosis units. The standard predates the PFAS regulatory wave by decades, it has historically been the certification of record for arsenic, hexavalent chromium, nitrate, fluoride, and TDS reduction. PFAS was added to the optional claims list with the same 1.5 µg/L challenge / 70 ng/L pass criterion as P473, so the headline number looks identical. The mechanism is not.
A carbon-block filter certified under NSF 53 P473 removes PFAS by adsorption: the long fluorinated tail of a PFOA or PFOS molecule sticks to activated carbon pore walls through hydrophobic interactions. The capacity is finite, the kinetics depend on contact time, and the medium eventually saturates, which is why P473 testing measures effluent across the entire rated cartridge life, not just at start-up. Short-chain PFAS (smaller, more polar molecules) are markedly harder to hold onto carbon.
An RO unit certified under NSF/ANSI 58 removes PFAS by membrane rejection. A semi-permeable thin-film composite membrane physically excludes molecules above a size/charge threshold that includes essentially all PFAS, long and short chain alike. Published peer-reviewed work (including the 2020 Duke / North Carolina State analysis of household filters across the Cape Fear basin) consistently puts reverse-osmosis rejection above 94% for the full PFAS panel, including HFPO-DA (GenX) and PFBS, which carbon-block filters often struggle with. That is the headline difference and the reason this site recommends RO for any household on a documented hotspot utility. The trade-offs, cost, and wastewater ratio are covered in the RO vs pitcher guide.
P473 was written for PFOA and PFOS specifically. Both are eight-carbon "long-chain" perfluoroalkyl acids, large hydrophobic molecules that activated carbon was always going to be reasonably good at holding. The story changed when industry transitioned away from PFOA in the early 2010s.
The replacements, HFPO-DA (Chemours' "GenX"), PFBS, PFBA, ADONA, and a sprawling list of others, are short-chain. Fewer carbons, smaller molecules, more polar, less affinity for hydrophobic adsorbents. Independent breakthrough studies on point-of-use carbon blocks have repeatedly shown that filters which knock PFOA from 100 ng/L down to non-detect leave half or more of the influent HFPO-DA in the effluent. The chemistry that makes carbon a good answer for PFOA makes it a mediocre answer for GenX.
P473 does not require the manufacturer to demonstrate any reduction of short-chain PFAS. A carbon-block filter can carry a fully valid NSF 53 P473 certification and still pass through significant fractions of HFPO-DA, PFBS, or PFBA. This is not a flaw in the test, it is doing exactly what the protocol scopes, but it is a critical limitation to understand when reading a label.
People sometimes hope NSF/ANSI 401 covers the gap. It does not. NSF/ANSI 401 lists fifteen specific "emerging" compounds, meprobamate, phenytoin, atenolol, carbamazepine, naproxen, ibuprofen, estrone, BPA, nonyl phenol, linuron, DEET, metolachlor, trimethoprim, TCEP, and TCPP. PFAS are not on the 401 list. A filter marked "NSF 401" reduces those fifteen molecules to defined thresholds; it makes no claim about PFOA, PFOS, HFPO-DA, or any other per- or polyfluoroalkyl substance. The right way to think about 401 is "tested for pharmaceutical residues and a handful of household-product contaminants", useful, but not a PFAS certification.
For short-chain PFAS coverage today, the practical answer is reverse osmosis certified to NSF/ANSI 58. Membrane rejection is largely indifferent to chain length within the PFAS family, the molecules are all big enough relative to the membrane cut-off. Households on utilities flagged for GenX, PFBS, or PFBA in UCMR 5 results should default to RO rather than a P473 carbon block, regardless of what the label promises.
Marketing pages are not the source of truth. A product page can say "NSF certified" and link to a generic NSF logo without the certification covering anything you care about. The source of truth is the NSF Certified Product Listings database at nsf.org/certified-products, which is maintained by the certifying body itself and lists every product with the standard it was tested under, the specific contaminants claimed, and the model and replacement-cartridge part numbers.
The lookup is straightforward. Search the manufacturer name, find the model, and read the "Contaminant Reduction Claims" panel. The phrase you are looking for is literally "PFOA" and "PFOS" in the claims column, with the standard cited as NSF/ANSI 53 (for adsorption media) or NSF/ANSI 58 (for RO). If those words are not in the listing, the product is not certified for PFAS, no matter what the box says.
Three traps to watch for. First, "tested to NSF standard X" is not the same as "certified to NSF standard X", only the latter means an accredited third party witnessed and validated the test. Second, the certification covers a specific replacement cartridge; an off-brand cartridge that fits the same housing is not certified and may behave entirely differently. Third, certifications are tied to a flow rate and rated capacity; using a P473-certified cartridge well past its rated litres will silently let breakthrough happen. The reduction claim is only valid through the rated cartridge life printed on the listing.
If you want a second opinion, two other accreditation bodies issue equivalent certifications against the same ANSI standards: the Water Quality Association (WQA) Gold Seal and IAPMO R&T. Both maintain public listings. A WQA Gold Seal with a PFOA/PFOS claim against NSF/ANSI 53 is, for practical purposes, equivalent to the NSF mark, same standard, different certifier.
Utility with low or non-detect PFAS in UCMR 5. Roughly nine out of ten US public water systems sampled under UCMR 5 returned non-detects across the regulated PFAS panel. If your utility is one of them, check the UCMR 5 explainer for the lookup, a NSF/ANSI 53 P473-certified carbon block at the kitchen tap is reasonable belt-and-braces protection. The pre-treatment burden is already low, so the gap between "carbon block to 70 ng/L" and "RO to non-detect" is academic.
Utility on a documented hotspot. If your zip falls inside one of the watersheds plotted on the hotspot map, the Cape Fear basin, the Merrimack Valley, parts of New Jersey, the Decatur-area Tennessee River, military installations across the West and the Mid-Atlantic, go straight to under-sink reverse osmosis certified to NSF/ANSI 58 with the PFOA/PFOS reduction claim. Carbon-block-only is not adequate where short-chain PFAS (HFPO-DA, PFBS, PFBA) are present, and most hotspots involve a short-chain component.
Private well. Wells are not covered by UCMR 5, the new MCLs, or any utility-side monitoring. If you draw from a private well within a few miles of an airport (AFFF firefighting foam), a military base, a chrome-plating facility, a textile or paper mill, a chemical manufacturer, or a documented PFAS plume, test first via a state-accredited lab, see the home PFAS testing labs guide for accredited options around $300, and treat the result as the basis for filter selection. Default to whole-house GAC + point-of-use RO if any of the six EPA-regulated PFAS come back above 4 ng/L.
In all three cases the underlying point is the same: the certification on the box must explicitly carry a PFOA/PFOS reduction claim, validated by NSF, WQA, or IAPMO against NSF/ANSI 53 P473 or NSF/ANSI 58. Everything else, "advanced filtration," "removes contaminants," "tested for over 200 substances," "ZeroWater," "alkalising," "structured", is noise. Pull the certified product listing, read the claims column, match the model number, replace the cartridge on schedule. That is the entire job.
Last reviewed 30 June 2026 by the How Safe Is My Water editorial desk. Corrections: /corrections.html.
State UCMR 5 baseline, hotspot match if applicable, the six regulated PFAS with their MCLs, a filtration matrix keyed to the certifications above, lab options, and a 14-day plan. Instant PDF. No subscription.
Pull my report, $15