NSF 42 vs 53 vs 401 vs 372 vs P473: what each standard means for a refrigerator filter

The short version

• 7 of the 45 filters we checked held NSF/ANSI 53, the standard that carries lead reduction, for the exact model sold.
• 5 of the 45 held NSF/ANSI 42 only, the taste, odor and chlorine standard.
• NSF/ANSI 372 is a materials standard, not a removal standard: Icepure’s RWF3600A and RWF1500A held 372 without the 42 their listings claimed alongside it.
• 2 of the 45 held NSF P473 for PFOA and PFOS, both GE OEM filters.
• 22 of the 45 listings claimed certifications the databases do not show for the model being sold.

NSF/ANSI 42 is the taste, odor and chlorine standard, and 5 of our 45 filters held it and nothing else

NSF/ANSI 42 is the aesthetic-effects standard. A filter certified to it has been tested to reduce chlorine taste and odor. Each listing records the reduction claim for the specific model, and for the 42-only filters in our sample the databases state chlorine reduction. A 42 certification says nothing about lead, PFAS, or pharmaceuticals.

Five filters in our 45-SKU sample were recorded as certified to NSF/ANSI 42 only: GlacierFresh GF-1000P, GlacierFresh GF-XWF-N, Mist’s EDR1RXD1 replacement, HMASAV’s W10295370A replacement, and Waterdrop WD-F38. All five are legitimately certified for the aesthetic standard. None of the five carries a 53 listing in the databases we searched.

This is the most common honest certification in the aftermarket. A 42-only filter is not a fake, and it is not uncertified. It is a filter certified for the thing most people actually notice: whether the water tastes like chlorine.

NSF/ANSI 53 is the health-contaminant standard, and only 7 of our 45 filters held it

NSF/ANSI 53 covers health-related contaminants including lead, and certification is recorded against a standard and a specific model. A 53 listing is therefore the right place to look for lead reduction, and it is also the place to check whether the specific model carries a lead claim.

The 7 filters in our sample confirmed certified to NSF/ANSI 53 for the exact model sold:

• Waterdrop WD-DA29-00020B (Samsung DA29-00020B replacement): NSF/ANSI 42, 53 and 401
• Waterdrop LT1000PC / WD-LT1000P (LG LT1000P replacement): NSF/ANSI 42, 53 and 401
• Waterdrop’s LT700P replacement: NSF/ANSI 42, 53 and 401, split across two element codes (WD-LT700P for 42; WDP-LT700P for 53 and 401)
• Maxblue MB-LT1000P (LG LT1000P replacement): NSF/ANSI 42, 53 and 401
• Filterlogic FL-RF46 (LG LT1000P replacement): NSF/ANSI 53 and 401, without 42
• GE XWFE (OEM): NSF/ANSI 42, 53, 372, 401 and P473, listed under “GE Appliances, A Haier Company”
• GE RPWFE (OEM): the same file set, NSF/ANSI 42, 53, 372, 401 and P473

Five of the seven trace to one contract manufacturer. Waterdrop’s three entries and Maxblue’s single entry are all listed under Qingdao Ecopure Filter Co., Ltd., and Filterlogic’s FL-RF46 is also an element on a Qingdao Ecopure listing. The only 53 certifications in our sample that do not trace to Qingdao Ecopure are GE’s two OEM filters. If you are shopping for 53, the verifiable set in our 45-SKU sample came down to one manufacturer’s product family plus two GE OEM filters.

NSF/ANSI 401 covers emerging compounds, and the same 7 filters held it

NSF/ANSI 401 covers a defined list of emerging compounds. We did not attempt to enumerate that list; what our dataset records is which models hold a 401 listing.

The 7 filters that held 401 are the same 7 that held 53: Waterdrop WD-DA29-00020B, Waterdrop LT1000PC / WD-LT1000P, Waterdrop’s LT700P replacement, Maxblue MB-LT1000P, Filterlogic FL-RF46, GE XWFE and GE RPWFE. We found no filter in the 45 that held one of the two standards without the other.

That overlap is the finding. In our data, 401 and 53 arrived together for every filter, which makes a listing that claims 401 without a verifiable 53 the pattern worth checking. Our sample contains one: PurePlus’s DA29-00020B replacement, whose listing claims “NSF/ANSI 401, 53 & NSF P473 Certified” and which could not be confirmed under any searched name in NSF, WQA or IAPMO.

NSF/ANSI 372 is a materials standard, and two Icepure listings treat it as if it were a removal claim

NSF/ANSI 372 certifies the lead content of the materials that contact drinking water. It is a lead-content standard for the product’s own housing and components, not a lead-removal performance standard. A filter certified to 372 has been verified not to introduce problematic lead from its own materials. It has not been tested to remove lead arriving in the water.

This is the most common conflation in the listings we checked, and it produces a repeatable failure. Icepure’s RWF3600A and RWF1500A listings both claim “NSF 42, 372 Certified”. Both are genuinely certified, under Icepure’s IAPMO registrant TOP PURE USA INC, for NSF/ANSI/CAN 372 only. The 42 half of the claim does not appear for either model. A reader who treats 372 as lead removal is reading a materials certification as a health-performance claim, and a reader who treats the pair as a bundle gets half of it.

The reverse case exists in the same brand. Icepure’s RWF5200A claims “NSF 42, 372 Certified” and holds both: NSF/ANSI 42-2021 and NSF/ANSI/CAN 372-2022, model-exact, under TOP PURE USA INC. Same brand, same claim style, different outcome per model. The model number matters more than the brand name, which is the pattern our 45-filter check found.

NSF P473 covers PFOA and PFOS, and only GE’s two OEM filters held it in our sample

NSF P473 is the protocol for PFOA and PFOS reduction: two compounds within the broad PFAS family. A filter certified to P473 has been tested for reduction of those two specific compounds, and the certification does not extend to the PFAS family as a whole.

In our sample, exactly two filters held P473: GE XWFE and GE RPWFE, both under the IAPMO registrant “GE Appliances, A Haier Company”. No aftermarket filter in the sample, including the five that held 53 and 401, carried a P473 listing for the model sold.

A 53 certification does not by itself put lead on that model’s reduction list

This is the distinction that separates a careful reading of a listing from a reassuring one. NSF/ANSI 53 is a standard with a set of contaminant reduction claims, and a manufacturer certifies a specific model against specific claims within it. A listing that says the model is certified to NSF/ANSI 53 tells you the model was tested to the standard. It does not tell you which claims under the standard the model holds, and lead is one claim among several.

The public databases resolve this. Each listing states the reduction claims for the specific model, not only the standard, which is why our verification recorded certification both by standard and by model-exact match. Our own dataset stops at the standard level for most rows, and we say so in the limits section below.

What to look for depends on what you are trying to remove

If you want better-tasting water, NSF/ANSI 42 is the standard to look for. It is the aesthetic standard and the most widely held certification in the market we checked.

If you want lead reduction, look for NSF/ANSI 53 with lead as a stated claim on the exact model. In our 45-SKU sample that narrowed to seven filters, five of them from one contract manufacturer. The independent lead-testing record is thinner than most buyers assume: the only substantial public lab program on fridge-style filters tested 32 known counterfeits in 2018, and all 32 failed the NSF/ANSI lead standard at twice rated life. No comparable lab program has tested the general aftermarket since. Our review of the independent lead-testing record sets out every study we could find and the ones that do not exist.

If you want PFAS reduction, look for NSF P473 on the exact model. Two filters in our sample held it, both OEM.

If you want materials assurance, meaning confidence that the filter’s own components do not add lead, NSF/ANSI 372 is the standard, and it is a real certification worth having. It is not a substitute for 53, and it is the standard most often misread.

Why a listing can name a standard it does not hold

Of the 45 listings we checked, 39 made or implied a certification claim. Of those 39, 22 (56%) claimed certifications the official databases did not show for the model being sold. Across all 45 listings, 22 overstate, 10 match exactly, 2 understate, 2 cite a different standard than the one held, 6 make no certification claim, and 3 could not be evaluated.

The most common pattern was not a fabricated standard but an extended halo: a brand holds a real certification for some models, and the listing applies that certification language to a model that is not on the certified list.

AquaCrest is the cleanest case in our set. The brand genuinely holds NSF listings, for exactly five models, all NSF/ANSI 42-only. None of the six AquaCrest SKUs in our sample is on that list, yet each claimed certification the database did not show for the model sold. The same shape covers 14 of the 45 filters: the brand is listed, and the model in the box is not.

The certification bodies publish their listings and the searches are free. NSF’s database, IAPMO R&T’s product directory and WQA’s product finder all accept a brand and model search, and each search takes under a minute. Our method and disclosure page explains how we ran all three. The short version: search the model number, not the brand name, and read the standard and the reduction claims together.

What we could not determine

Our 45-SKU sample is a best-seller snapshot from August 2026, not the whole market. The counts above (5 at 42-only, 7 at 53, 7 at 401, 2 at P473) describe those 45 filters on that date, and certifications change: listings are added, models are de-listed, and a filter we could not verify then may be certified now.

Seven of the 45 filters could not be verified at all, meaning the entity behind them was absent from all three databases under every name we tried. We record those separately from the nine filters with no listing anywhere and the 14 whose brand is listed but whose model is not. “Could not verify” is not the same as “uncertified”, and we keep the two categories apart in every row of the dataset.

Two structural limits belong in plain language. First, certification databases index by certificate-holder name, so a filter genuinely certified under an unlisted contract manufacturer’s name would be missed by our protocol and recorded as unverifiable rather than uncertified. Second, we recorded the standards held for each model, not the full per-model reduction-claim matrix. A model can hold 53 without holding the lead claim within it, and our data does not resolve that at claim level for all 45 rows.

Two of the 42-only rows also rest on a brand model family rather than a single cross-walked element code: Mist’s CWMF series and HMASAV’s HM-prefixed models. We recorded those as certified for 42 on the strength of the brand’s listing, with the caveat that the exact retail SKU was not cross-walked to one code.

We could not confirm 372 for every filter in the set either. It is often co-listed with 42 or 53 and sometimes appears alone, and the absence of a 372 listing is not evidence that a filter leaches lead. It means the certification could not be traced.

Where to check your specific filter next

The full 45-row dataset, with every source URL and capture date, is in our verification study of the best-selling filters. The pass that produced these counts did not resolve every row, and that page states which ones and why.

If you know your fridge model, the fitment index maps 50 refrigerator models to the filter they take and to the SKUs in our dataset, so you can see whether the compatible filter sold for your fridge is one of the verifiable ones or one of the fourteen whose brand is listed but whose model is not. Our method page covers how the three databases were searched and how the affiliate side of the site is funded.

If your question is lead removal specifically, the independent lead-testing record is the page that answers it directly. This page explains what a 53 certification does and does not promise; that one sets out what independent labs have actually measured.