How to check whether a water filter is actually NSF certified

The short version

• Of the 45 best-selling refrigerator filters we checked in August 2026, 15 (33%) were confirmable as certified for the exact model sold; 30 (67%) were not.
• 14 of 45 (31%) came from brands holding real certifications that did not cover the model being sold.
• 22 of the 39 listings making or implying a certification claim (56%) claimed certifications the databases did not show.
• Nine of 45 (20%) had no certification listing anywhere; seven (16%) could not be matched to a record for that exact model in any of the three databases, even after a clean re-check.
• GE’s XWFE and RPWFE are certified in IAPMO, not in NSF. A buyer checking only NSF would have concluded they were uncertified.

The model number is the check, and the brand is not

Certification is issued to a manufacturer for specific models, not to a brand in general. That single fact produces most of the confusion we measure. In our 45-filter sample, 14 filters (31%) came from brands holding real certifications for other models.

AquaCrest is the clearest case in our data. The brand genuinely holds NSF listings for exactly five models: EFF-6008S1/AQF-FF08, EFF-6032S/AQF-FF32D, EFF-6038S1/AQF-FF38, AQF-FF12 and EFF-6002S/AQF-FF15, all certified to NSF/ANSI 42 only. None of the AquaCrest SKUs in our sample matched any of them. Waterspecialist’s NSF footprint is two models. Pureza’s is three (PF34, PF54, PF55). Crystala’s is 44 products across three manufacturers, all NSF/ANSI 42 only, and neither of the Crystala filters we checked (a DA97-17376B replacement and an XWFE replacement with a chip) was among them. A brand page reading “NSF certified” and a database listing five unrelated models can both be true at once.

The table above shows all 45 filters by verification result. Fifteen were certified for the exact model sold. Fourteen were brand-listed but model-absent. Nine had no certification listing anywhere. Seven could not be verified at all. That last split matters: a brand holding real certifications for other models is a different finding from a company that cannot be located in any of the three databases, and the second tells you less, not more.

Across all 45 SKUs, the claims on the listings compared to the databases as follows: 22 overstated (49%), 10 matched (22%), two understated (4%), two cited a different standard than the one actually held (4%), six made no certification claim (13%), and three could not be evaluated (7%).

Three databases exist, and checking only NSF will mislead you

NSF’s certified drinking-water treatment unit listings are the largest single record of filtration-performance certification, and for most aftermarket filters they are enough. They are not enough for everything.

GE’s two OEM filters, XWFE and RPWFE, return no match in NSF under any name we tried: a TradeName=GE search returns 257 rows tagged GE but none of them XWFE or RPWFE, and Company=General Electric, Company=GE Appliances and TradeName=GE Appliances each came back with zero results. Their certification sits with IAPMO under the certificate-holder name “GE Appliances, A Haier Company”, and it is broad. XWFE is covered by W-9755 for NSF/ANSI 42-2022, W-9756 for NSF/ANSI 53-2021, W-9757 for NSF/ANSI 372-2020, W-9758 for NSF/ANSI 401-2022 and W-11362 for NSF P473-2016 (PFOA and PFOS). RPWFE matches the same five files plus W-9975 for CSA B483.1. These are among the most thoroughly certified products in our entire sample, and a search of NSF’s database alone would have shown nothing.

Samsung runs the opposite way. Samsung Co., Ltd. is a WQA-certified manufacturer for NSF/ANSI 42, 53, 372 and 401, but only for four models: HAF-HIM-1, HAF-HIM-2, RWP70a15def and RWP90. DA29-00020B / HAF-CIN/EXP, the Samsung OEM filter in our sample, is not one of them. Samsung Electronics separately holds two active IAPMO files, W-14267 (NSF/ANSI/CAN 61-2023, drinking water system components) and W-6572 (Lead Free Plumbing Products), and neither is a filtration-performance certification. The honest reading is narrow: the company is certified, the model in front of you is not confirmed.

Two more routes worth knowing. Waterdrop’s certified models appear in NSF under its manufacturer, Qingdao Ecopure Filter Co., Ltd., including the exact model WD-DA29-00020B, certified for NSF/ANSI 42, 53 and 401, with the variant codes WDS-DA2900020B and WDP-DA2900020B on the same listing. Icepure runs through IAPMO instead: RWF5200A appears under the registrant TOP PURE USA INC as WP-12585 (NSF/ANSI 42-2021) and WP-12587 (NSF/ANSI/CAN 372-2022), and nowhere in NSF. The certifying bodies are not interchangeable, and the entity holding the certificate is often not the brand on the box.

Step by step: running the three searches

Each database asks a slightly different question and each has its own failure mode. This is the order we use.

1. Write down the exact model code before you start. Copy it from the product page or the cartridge, including suffix letters, and note any alternate codes the listing gives. A filter sold as LT1000P is the same product as ADQ74793501. A filter sold as AQF-FF32 may not be the same product as AQF-FF32D.
2. Search NSF. Go to info.nsf.org, open the Drinking Water Treatment Units listings, and search twice: once by TradeName (the retail brand) and once by Company (the manufacturer or certificate holder). A real match names the model, the manufacturer and the standard numbers. NSF’s search was the most reliable of the three across our whole pass.
3. Search WQA. Go to find.wqa.org and use the keyword route directly, at find.wqa.org/find-products#/keyword/your-model, rather than filling in the on-screen form. In our research pass the direct keyword URL was the more reliable route.
4. Search IAPMO R&T. Go to pld.iapmo.org and search both the Listee Name field and the Listed Model field. This directory is a JavaScript-driven search form and it is the one that gave us trouble: when more than one browser session was open against it, typed terms were overwritten mid-search. Search one term at a time, let the page settle, and reload the page between terms instead of clearing the field.
5. Repeat all three searches using the certificate holder’s name. This is where a large share of real matches live, and it has its own section below.
6. Record what you found before you decide anything. Model, standard numbers, certificate-holder name, file or certificate number, and the date. Certification records change.
7. Treat any unrecognised name in the results as unconfirmed until you verify it is the same company. A keyword match is not a model match. WQA’s record for the name “Great Value” belongs to MAVEA LLC and covers pitcher filters, not Walmart’s refrigerator filter.

The certificate holder is a different search term from the brand

If a search of the retail brand comes up empty, the brand is the wrong term, not necessarily the wrong product. Certificates sit with whoever holds the registration, and that name changes with corporate history and with manufacturing.

• GE Appliances, A Haier Company is GE Appliances’ registered name since the 2016 sale to Haier. Both GE OEM filters are listed under it, in IAPMO.
• Electrolux Home Products is Frigidaire’s parent, holding five active IAPMO certificates: NSF/ANSI 42-2022 (WP-14065), NSF/ANSI/CAN 372-2022 (WP-14066), NSF/ANSI 53-2021 (WP-16165), NSF/ANSI 401-2023 (WP-30821) and CSA B483.1 (WP-30822). The WF3CB model itself still does not appear in that listing, which is why we classify that filter as brand-certified but model-unconfirmed rather than certified.
• Samsung Electronics Co., Ltd. and Samsung Co., Ltd. are two separate registrations with two different sets of standards. Searching “Samsung” without knowing which one you want produces a confusing mixture of both.
• Qingdao Ecopure Filter Co., Ltd. is the manufacturer behind Waterdrop’s certified models, and it is also the entity named in NSF listings for other brands’ filters using the same model numbers.
• TOP PURE USA INC is the IAPMO registrant behind the Icepure listings, with a different certificate set per model. RWF5200A holds NSF/ANSI 42 and 372. RWF3600A and RWF1500A hold NSF/ANSI/CAN 372 only. One company, three different answers.
• Ningbo Koze Purification Equipment Co., Ltd. is the manufacturer behind GlacierFresh’s NSF/ANSI 42 listings. The Crystala trade name spans three manufacturers — Ningbo Koze, Ningbo Pureza Technology LLC and YiHangGou Trading, all also under NSF/ANSI 42.
• Shenzhen Jiquanlin Water Purification Technology Co., Ltd. is the manufacturer behind HMASAV.

Two warnings come out of the same exercise. A name match is not a model match. And a familiar word can belong to a different business: “Pureline Treatment Systems, LLC” is a real WQA entity, but its listings are for chemical-generator products (NSF/ANSI/CAN 372, 60 and 61), not refrigerator filter cartridges. Check the product category and the standard, not just the name.

One changed suffix is a different listing

Near-miss codes are the most common way a search turns up something that looks like an answer but is not.

• AquaCrest’s AQF-FF32. The exact code without the D suffix does not appear as a distinct NSF entry. AQF-FF32D does, as an element code under Waterdrop’s listing EFF-6032S, certified to NSF/ANSI 42 only.
• GlacierFresh’s GF-XWF and GF-XWF-N are listed under Ningbo Koze for NSF/ANSI 42. The RPWFE-compatible GFRPWFE is not in that set.
• Waterdrop’s LT700P line splits by element code. WD-LT700P is listed for NSF/ANSI 42 only, while WDP-LT700P carries 53 and 401, both under the same listing, EFF-6032S. Waterdrop’s LT1000P element, WD-LT1000P, doesn’t split the same way — it’s listed under EFF-6046S for all three standards, 42, 53 and 401, together. WD-F38 and WD-F38-AL sit under EFF-6038S1 with NSF/ANSI 42 only, and 372 does not appear for that element even though the product listing claims it.
• The same Waterdrop Samsung listing carries WDS-DA2900020B and WDP-DA2900020B alongside WD-DA29-00020B.

Practical rule: search the code you have, then search the base code with the suffix removed, and treat any difference in letters as a different row until the database says otherwise.

A certificate number on a listing proves less than it appears to

HMASAV’s listing cites “NSF 42 certification (Certificate C0791840, NSF International, 2023)”. That is the most checkable claim in our entire sample, and it still could not be confirmed. We verified that HMASAV is a real NSF trade name, that the manufacturer is Shenzhen Jiquanlin Water Purification Technology Co., Ltd., and that 32 HM-prefixed models are certified to NSF/ANSI 42 only. We could not cross-walk W10295370A to a specific HM element code, and NSF’s public search exposes no certificate-number lookup at all, so C0791840 is not independently verifiable from outside.

That is the limit of a certificate number. It is useful when the certifier lets you search it and decorative when they do not. If a number is the only verifiable-looking detail on a page, the next step is to contact the certifier directly and quote it.

Reading the standard numbers you find

Standard numbers are not interchangeable, and a listing naming one is not making the same claim as a listing naming another.

• NSF/ANSI 42 covers aesthetic effects: chlorine taste and odour, and particulate reduction.
• NSF/ANSI 53 covers health effects, including lead and cyst reduction.
• NSF/ANSI 372 covers the lead content of the materials the filter is built from, not what it removes from water.
• NSF/ANSI 401 covers emerging contaminants including pharmaceuticals.
• NSF P473 covers PFOA and PFOS.

Two patterns are worth knowing before you read any claim. Certification splits three ways among the 15 filters we confirmed: five for 42 alone, five for 42 and 53 together, and five for other combinations of standards. And certifications come in sets that do not always include the baseline. Filterlogic’s FL-RF46 is certified for NSF/ANSI 53 and 401 but not 42, the only SKU in our 45 with that shape. Waterdrop’s LT1000P element carries 42, 53 and 401, with 372 absent. If lead removal is what you care about, the standard to look for is 53. The full comparison is in NSF 42 vs 53 vs 401 for refrigerator filters.

Nothing in the public record is not the same as uncertified

A negative search is a fact about the record, not a verdict on the filter. Keep three states apart. One: the listing makes no certification claim at all, and the database is also empty. Cenmela’s CML006 and Haprait’s HP006 are both in that state, and we record the absence of a claim and the absence of a listing as two separate facts. Two: the brand is certified but the model is not listed, which covers 14 of our 45. Three: neither the brand nor the model could be matched to a record in any of the three databases for that specific SKU, which covers 7 of 45.

That third group includes both LG OEM filters. Nothing appears under TradeName=LG or Company=LG in NSF, no LG or LG Electronics listee appears in IAPMO, and WQA returns no results for LT1000P or LT700P. The model numbers themselves appear only under third-party compatible-filter makers.

A related but separately recorded case — no listing found, rather than could-not-verify, is Whirlpool’s EDR1RXD1: no listing under Whirlpool or everydrop in any of the three databases, and the EDR1RXD1 model number maps only to Qingdao Ecopure Filter Co., Ltd., a compatible-filter maker rather than Whirlpool. In NSF, the name “Whirlpool” returns water softeners (NSF/ANSI 44) and reverse-osmosis systems (NSF/ANSI 58), not this cartridge. That pattern is consistent with a certificate held under a name that the databases do not cross-reference to the retail brand, and we label it that way.

Two things we do not do. We never label a filter uncertified; the sharpest status we use, brand-listed-model-absent, applies only when a database shows a complete model list for a brand and the specific SKU is missing from it. And we do not call a filter counterfeit without a named source using that word: the AHAM study that tested 32 counterfeit filters, all 32 failing the NSF/ANSI lead standard at twice rated life, studied known counterfeits rather than the ordinary aftermarket. “Could not verify” and “uncertified” stay separate in every row we publish.

What we could not determine

• All checks are point-in-time. Every certification result in this guide was captured on 7 August 2026. Databases and best-seller lists change.
• Certificate-holder indexing is the main structural limit. A filter certified under an unlisted contract manufacturer’s name would be missed. Those rows are flagged as unverifiable, never as uncertified.
• One search tool is fragile. IAPMO’s directory is a JavaScript-driven form. Our original pass ran five browser sessions in parallel against it and produced cross-contaminated results, which a later single-session re-check corrected. A residual 16% genuinely remained unresolved after a clean attempt.
• No performance testing was done for this guide. This is a records check, not a lab test. The lead-removal testing record, including what exists and what does not, is in the independent lead-removal testing record.
• Some model-to-code links could not be resolved. HMASAV’s W10295370A and Mist’s EDR1RXD1-branded product were not cross-walked to specific element codes. Mist’s 21 CWMF models (CWMF001 to CWMF052) are NSF/ANSI 42 only, but we cannot say which one is in a given box.
• Private-label products are opaque. The contract manufacturer behind Walmart’s GE XWFE-compatible filter could not be identified from the public listing, so that row stays could-not-verify rather than being judged either way.

Our full method, classification rules and affiliate disclosure are at /about/.

Where to check next

If you have one filter in front of you, the platform guide holds the full verification record: Samsung, LG, Whirlpool everydrop, GE XWFE and RPWFE, Waterdrop and AquaCrest. If you are trying to work out whether a claim is a certification claim in the first place, read certified vs meets NSF standards next. If you need to know which cartridge your fridge takes before you search for it, the fitment index covers 50 models at /filters/. Every figure above comes from the 45-filter dataset behind our verification study.