Star Water Filter®
Updated: 20 August 2026

TECHNICAL COMPARISON OF GRAVITY WATER FILTERS

Gravity-fed water filter comparison: what criteria should you consider when making an informed choice?

To compare a gravity water filter seriously, several criteria are particularly useful: the filtration architecture, published independent laboratory analyses, the volumes actually tested, the stability of performance over time, the exact scope of the certifications and the filtration flow rate in real-world use.

A gravity water filter is a standalone system, not connected to the water supply, in which water passes through the filter elements under the force of gravity. Its performance therefore cannot be assessed solely on the basis of flow rate, claimed lifespan, or brand recognition.

Among the 11 systems reviewed in this comparison and according to the documentary criteria used, Star Water Filter® presents, as of the date of this update, the most comprehensive body of documentation among the brands analysed in terms of documented filtration architecture, published independent analyses, and monitoring of performance over time. This conclusion does not constitute a universal ranking: price, intended use, system capacity, or other criteria may lead to a different choice.

NSF®-certified filter Exact scope documented
Independent laboratory analyses Published results
Documented stability Long-term reading
REACH compliance Material transparency

QUICK READ

Key takeaways

01

A claimed lifespan does not show up to what volume the filter’s performance has actually been tested.

02

A result obtained with a new filter does not demonstrate the stability of performance over time.

03

A certification is not sufficient on its own: you need to identify precisely what is certified.

04

When comparing filters, contaminants, protocols, volumes and test conditions must be considered in context.

INTRODUCTION

Comparing gravity water filters requires more than a promise or a logo

Most comparisons of gravity water filters compare brands, announced service life or certification claims without clearly distinguishing what comes from an independent analysis, a normative framework or simple manufacturer-published information.

This page takes a different approach: comparing the evidence that is actually accessible. We selected documentary and technical criteria that help read the differences between systems more seriously: filter architecture, the presence of a nanofiltration membrane, documented long-term stability, the depth of published analyses, filtration flow rate, NSF® certification and REACH compliance.

Eleven gravity-filtration systems or configurations were reviewed: Star Water Filter® / Ultimate Star Filter®, Berkey / Black Berkey®, Coldstream FTO Plus, British Berkefeld / Ultra Sterasyl, Weeplow O’Pure 2, Euroguard Classic, Phoenix KokoCarb, Purewell K Series Classic (K8626 / K8012), Katadyn Gravidyn, Ecofiltro and Orinko. The aim is not to assign an arbitrary score, but to help compare what is documented, what remains partial, what is only claimed, and above all the actual volume up to which performance has been tested.

This page compares gravity-fed water filters only. If you are still deciding between a filter jug, tap-mounted filter, under-sink system, reverse osmosis and gravity-fed filtration, first consult our comparison of the main water filtration solutions. The rest of this page explains how to distinguish between the levels of evidence available for gravity-fed systems.

METHODOLOGY

Not all comparisons rely on the same level of proof

Many comparisons pit brands, prices or claims against one another without clearly distinguishing the nature of the available evidence. Here, we first compare the level of documentary evidence.

Scope of the analysis

Comparison updated on 20 August 2026. Eleven systems or configurations were analysed: Star Water Filter® / Ultimate Star Filter®, Berkey / Black Berkey®, Coldstream FTO Plus, British Berkefeld / Ultra Sterasyl, Weeplow O’Pure 2, Euroguard Classic, Phoenix KokoCarb, Purewell K Series Classic (K8626 / K8012), Katadyn Gravidyn, Ecofiltro and Orinko.

The information is classified according to its nature: measured result, certification or compliance, published technical characteristic, manufacturer claim or absence of clearly identified public evidence.

The absence of clearly identified public evidence does not mean that a product is ineffective; it only means that, as of the date of review, we did not identify sufficiently clear public data to substantiate the point concerned.

To make the comparison easy to understand, we indicate the actual tested volume in litres whenever it is publicly available. A test on a new filter or over only a few litres does not provide the same information as a result verified after several hundred or several thousand litres.

For the flow rate, we use, wherever possible, a standard configuration with two filter elements, a very common setup among the gravity systems compared. When the design is different, or the manufacturer publishes only a value per cartridge or per system, we keep that value as published and state it clearly. When no clear numerical value is published, we simply state “no public numerical value”.

Hierarchy of levels of evidence in water filtration
01

Scientific proof

Analyses performed by independent laboratories accredited to ISO/IEC 17025, with published results and identifiable protocols.

→ Highest level of proof

02

Certification and compliance

A NSF®/ANSI certification provides evidence within the exact scope it covers. REACH compliance provides information about materials and substances, not filtration performance.

→ Always check what is actually certified

03

Public information

Data from manufacturers (technical sheets, product pages), without accessible independent validation.

→ Lowest level of proof

Not all information available on the market carries the same weight. This comparison prioritises the most documented and verifiable evidence, particularly regarding the filter architecture and stability over time.

COMPARISON TABLE

Compare gravity water filters using evidence and the actual volumes published

This table compares information publicly available as of 20 August 2026. When the tested volume is known, it is shown in litres so that the claimed service life can be compared with what was actually measured. Evidence at 5 L, 100 L, 1,500 L or 3,000 L does not provide the same information about how a filter performs over time. Flow rates are the values published by manufacturers for the stated configurations; they may vary with water level, water quality, clogging and the number of cartridges. The absence of identified public evidence does not mean that a product is ineffective: it means that we did not find sufficiently clear public data on the point concerned.

Technical reading
Present

Technical characteristic identified in the system.

Not present

Characteristic not identified in the system as of the review date.

Claimed / partial

Characteristic claimed or documented with limited scope.

Level of proof
Published proof

Accessible and verifiable data; the tested volume is stated when known.

Partial proof

Real data, but limited by the volume, protocol, contaminant or model concerned.

No accessible proof

No sufficiently clear public data identified as of the review date.

On mobile, scroll the table horizontally to compare all systems.

To make reading easier, the table is divided into three blocks. Each block keeps the brands in columns and has its own horizontal scrollbar.

1. Filter architecture and design

Scroll this block horizontally to compare all brands.

Compared criterion Star Water Filter® Berkey / Black Berkey® Coldstream FTO+ British Berkefeld / Ultra Sterasyl Weeplow O’Pure 2 Euroguard Classic Phoenix KokoCarb Purewell K Series Katadyn Gravidyn Ecofiltro Orinko
Dominant filter architecture Multi-layer compositeActivated carbons, dedicated heavy-metals media, nanofiltration membrane and colloidal silver Proprietary composite mediaMedia blend claimed; detailed composition not published Ceramic + activated carbonCeramic shell + granular activated carbon Ceramic + activated carbonSterasyl® microfilter + granular activated carbon + lead-reduction media Activated carbon + 0.02 µm UFHollow-fibre ultrafiltration membrane; no nanofiltration Activated-carbon blockEuroguard Classic: compressed activated-carbon block Activated carbon + silverSilver-impregnated coconut-shell activated-carbon block Activated carbon + 0.01 µm UFK8012: activated carbon, silver ions and hollow-fibre UF membrane Silver-treated ceramic + activated carbonGravidyn cartridge for the Drip system Ceramic + activated carbon + silverClay/ceramic, activated carbon and colloidal silver Mainly activated carbonCoconut-shell activated carbon; little detail on any additional architecture
Detailed architecture beyond the main material Publicly documentedMulti-layer architecture explained and cross-referenced with published analyses Partially documentedProprietary media: mechanisms and full composition only partly detailed DocumentedCeramic and carbon clearly identified; relatively simple architecture DocumentedCeramic, carbon and lead-reduction media published; performance must be read criterion by criterion DocumentedCarbon + UF clearly claimed; finer composition less detailed Simple architectureThe Classic is clearly described as a compressed activated-carbon block Partially documentedCoconut carbon + silver published; little detail on the internal architecture beyond that Published architectureK8012 three-stage configuration clearly described by Purewell Published compositionSilver-treated microporous ceramic + high-performance activated carbon Published compositionThree materials clearly stated Limited detailActivated carbon stated, with little detail on any additional architecture
NSF®-certified activated carbons as filter media Documented internallySupplier certificates verified and held by SWF; not published to protect the formulation and suppliers Not identifiedNo public certificate for the activated carbon itself identified Not identifiedActivated carbon disclosed; NSF® certificate for the media not identified Not identifiedProduct/cartridge certifications are covered separately below Not identifiedTests according to certain standards must be distinguished from certification of the carbon itself No accessible proofA carbon block is listed under NSF® 42; this does not document the raw carbon itself as a certified filter media No accessible proofThe PCC8/KokoCarb cartridge is NSF® listed; this is not a certificate for the raw carbon itself No accessible proofNSF® documents are available for certain products/components; no public certificate identified for K8012 carbon as filter media Not identifiedActivated carbon disclosed; NSF® certificate for the media not identified Not identifiedActivated carbon disclosed; NSF® certificate for the media not identified Not identifiedActivated carbon disclosed; NSF® certificate for the media not identified
Nanofiltration membrane PresentIntegrated nanofiltration membrane Not present Not present Not present Not present0.02 µm ultrafiltration, not nanofiltration Not present Not present Not present0.01 µm ultrafiltration, not nanofiltration Not present Not present Not present
Bacteriostatic silver treatment PresentIntegrated colloidal silver Not publicly documented PresentSilver incorporated into the ceramic according to the filter documentation PresentSilver used in the ceramic formulation Not documented Not documentedThe term “ionised carbon” is not sufficient to establish the presence of silver PresentSilver-impregnated KokoCarb cartridge PresentSilver ions stated in the K8012 composition PresentSilver incorporated into the ceramic PresentColloidal silver Not documented
Published filtration flow rate / compared configuration 12 to 16 L/hWith 2 new Ultimate Star Filter® elements; around 4 L/h at about 3,000 L. Flow-rate evolution is published over use. Up to 13.3 L/hBig Berkey® 8.5 L with 2 Black Berkey® elements, upper chamber full; flow rate decreases as the water level falls. 1 L/h per filterPublished minimum flow rate for one CF163W; no extrapolation to a pair Up to 1.7 L/hBritish Berkefeld 8.5 L system with 2 Ultra Sterasyl® cartridges; published manufacturer value 3.8 L/hWith 2 O’Pure 2 elements, value published by Weeplow. No public numerical valueNo clear numerical value identified for 2 Euroguard Classic cartridges on the public pages reviewed. Up to 16 L/hWith 2 KokoCarb cartridges, advertised commercial flow rate. Up to 15.1 L/hK Series K8626 with 2 K8012 filters: up to 4 gal/h claimed 4 L/hPublished value for the Gravidyn cartridge; not extrapolated to a pair. 1 to 2 L/hEcofiltro 20 L, single-piece ceramic system. 8 L/hWith 2 cartridges; 4 L/h claimed per cartridge.
Claimed service life / actual tested volume 6,000 L / pair3,000 L per filter; chemistry tested up to 3,200 L per filter, microbiology up to 3,000 L per filter 3,000 gal / element≈ 11,356 L claimed; tests published by contaminant, without demonstration up to that volume 1,500 L / filter or 6 monthsMany contaminants are tested up to 3,000 L per CF163W filter, beyond the nominal capacity 6 monthsNo nominal volume in litres published for Ultra Sterasyl; several performance figures are published without a stated cumulative volume 22,000 L per pairBacteria tested up to 300 L; PFAS, viruses, pesticides and metals mainly tested at start-up 3,000 gal claimed≈ 11,356 L; chlorine tested up to 300 L, metals up to 50 L, detailed PFAS results up to 35 L 20,000 L per pair / 12 monthsPFAS tested up to 1,500 L; chlorine/taste/odour NSF® certified up to 4,000 L 6,000 gal per pair ≈ 22,700 LCurrent K8012 capacity; public reports identified over 3 to 5 L and on an older K8627 system, not the current K8626 6 monthsNo nominal capacity in litres published; microbiological tests are cited without a stated filtered volume on the public datasheet 2 yearsNo clearly published nominal volume; PFAS tested on 1 L and other tests are mainly one-off 11,350 L / cartridgeMicroplastics tested on 1 L; PFAS test volume not publicly stated; no evidence up to 11,350 L

2. Stability, protocols and certifications

Scroll this block horizontally to compare all brands.

Compared criterion Star Water Filter® Berkey / Black Berkey® Coldstream FTO+ British Berkefeld / Ultra Sterasyl Weeplow O’Pure 2 Euroguard Classic Phoenix KokoCarb Purewell K Series Katadyn Gravidyn Ecofiltro Orinko
Documented long-term stability / end-of-life Published proofChemistry up to 3,200 L per filter; microbiology up to 3,000 L per filter, for a nominal service life of 3,000 L per filter Partial proofBroad test file, but the identified tests do not track performance up to the ≈ 11,356 L claimed per element Published proofMany criteria measured up to 3,000 L per CF163W filter; some performance figures decline significantly with volume Partial proofRecommended replacement after 6 months; cumulative volumes often not stated in public data Very partialBacteria up to 300 L; other families mainly tested at start-up versus 22,000 L claimed per pair Partial proofChlorine 300 L, metals 50 L, detailed PFAS results 35 L versus ≈ 11,356 L claimed Partial proofPFAS up to 1,500 L and NSF® chlorine up to 4,000 L versus 20,000 L claimed per pair Not demonstrated≈ 22,700 L claimed per pair; public reports identified over 3 to 5 L and on an older K8627 system Not publicly quantifiableMicrobiological tests cited, but filtered volume not stated; replacement recommended after 6 months Not demonstrated over timeClaimed service life of 2 years; recent tests but mainly one-off Not demonstrated up to 11,350 LRecent analyses published, but no public follow-up up to the claimed capacity
Published independent analyses with readable protocol and volumes Published proofIndependent reports, readable protocols and volumes; several measurement points up to 3,000 / 3,200 L per filter Published evidence, but scatteredSeveral independent reports exist; some volumes are low or do not track performance up to the claimed service life Published proofDetailed datasheet, laboratory and volumes clearly published up to 3,000 L per CF163W filter Published proofTest data and certifications accessible, but cumulative volume is not always stated Partial proofReports accessible; 300 L for bacteria, several other tests at start-up Partial proofIAPMO reports accessible, but tests target volumes far below the claimed service life Partial proofReports accessible, with volumes varying by contaminant Partial proofThird-party reports available, but dated 2020 and relating to a K8627 system; correspondence with the current K8626 is not established Partial proofIndependent laboratories cited by Katadyn; full reports and volumes not published on the datasheet reviewed Published evidence, but one-offSeveral recent reports accessible; little litre-by-litre follow-up over use Published evidence, but one-offPublished French laboratory analyses; volumes of some tests not stated
Analyses according to standards / accredited laboratory Published proofISO/IEC 17025 laboratories; NSF®/ANSI protocols or other methods identified by contaminant Partial proofSeveral independent laboratories and normative references, but an older and heterogeneous test file Published proofIAPMO R&T; NSF/ANSI 42, 53 and P231; ISO 17025-compliant laboratory Published normative evidenceNSF®/ANSI 42, 53 and 401 for specific performance claims; NSF/ANSI/CAN 372 for lead content Partial proofSGS / other laboratories; some tests reference NSF®/ANSI or EPA, others use internal methods Partial proofIAPMO; tests according to NSF®/ANSI 42/53 on certain criteria, at low volumes Partial proofLaboratories and NSF®, EPA, APHA or AOAC methods depending on contaminant Partial proofPublic laboratory reports, but old and not clearly linked to the current K8626 / K8012 Partial proofBachema cited as an independent laboratory; methods and volumes not detailed on the public datasheet Partial proofMeasurlabs ISO/IEC 17025 for certain tests; other laboratories depending on contaminant family Partial proofIndependent French laboratories, including analyses stated as performed under COFRAC accreditation according to published documents
Heavy-metals analyses under NSF® 53 conditions (pH 6.5 / 8.5) Published proofSeveral metals tested at pH 6.5 and 8.5, with measurements up to 3,200 L per filter Published proofMetal tests exist at pH 6.5 and 8.5; no public follow-up up to 3,000 gal Partial proofMetals tested according to NSF®/ANSI 53 up to 3,000 L per CF163W filter; no dual pH 6.5 / 8.5 reading identified Partial proofLead published at pH 6.5; no dual pH 6.5 / 8.5 reading for Ultra Sterasyl Partial proofThirteen metals tested at start-up; some NSF®/ANSI 53 references Partial proofLead, cadmium and mercury tested up to 50 L, at approximately pH 6.5 only Partial proofMetals published, but pH conditions not equivalent to a full 6.5 / 8.5 reading Partial proofLead and mercury tested over about 3 L in a 2020 K8627 report; no NSF® 53 protocol at both pH values identified No comparable evidence identifiedNo public heavy-metals file under pH 6.5 / 8.5 conditions Partial proofMetals tested, without a public protocol equivalent to NSF® 53 at both pH values Partial proofPublished metal analyses, but no public protocol equivalent at both pH values
NSF® / ANSI certification: exact nature of the certification NSF®-listed filterUltimate Star Filter® USF-SWF-001: NSF/ANSI 42, material requirements only; lead-content requirements also assessed Black Berkey® not NSF® listedThe Berkey systems currently NSF® listed use Phoenix elements, which are different from the Black Berkey® elements compared here Tests according to standards; no certification identifiedCF163W tested according to NSF/ANSI 42, 53 and P231; no NSF® product listing identified Published proofUltra Sterasyl: NSF/ANSI 42 particulates, NSF/ANSI 53 cysts/turbidity, NSF/ANSI 401 microplastics, NSF/ANSI/CAN 372 lead content No product certification identifiedTests according to certain standards must be distinguished from NSF® certification NSF®-listed componentEuroGuard carbon block: NSF/ANSI 42, material and structural-integrity requirements only Published proofPCC8/KokoCarb: NSF/ANSI 42 chlorine / taste / odour up to 4,000 L; Phoenix systems also listed NSF® documents availableNSF/ANSI 42 and 372 on certain products/components; exact scope must be tied to the specific model and component concerned No NSF® certification identifiedFor the Gravidyn cartridge compared No NSF® certification identifiedLaboratory reports are published, but no NSF® listing identified Published proofOrinko cartridges: NSF/ANSI 42 for material safety only; NSF/ANSI/CAN 372 for lead content
Published material compliance (REACH, RoHS, NSF 372 or equivalent) Published proofREACH published: European compliance regarding chemical substances in materials; USF-SWF-001 filter NSF® listed for material requirements Not identified for Black Berkey®Current 372 listings concern systems fitted with Phoenix filters No public evidence identified Published proofNSF/ANSI/CAN 372 on listed Ultra Sterasyl systems/cartridges No public evidence identified Partially documentedNSF/ANSI 42 for the carbon block on materials/integrity; NSF 372 claimed by the manufacturer Published proofNSF® material requirements and lead content documented in Phoenix listings Documents availableNSF/ANSI/CAN 372 assessments and material documents published for certain components/systems; exact coverage of the K8626 must be checked No public evidence identified No public evidence identified Published proofNSF/ANSI 42 material safety + NSF/ANSI/CAN 372 lead content

3. Analysed contaminants and overall reading

Scroll this block horizontally to compare all brands.

Compared criterion Star Water Filter® Berkey / Black Berkey® Coldstream FTO+ British Berkefeld / Ultra Sterasyl Weeplow O’Pure 2 Euroguard Classic Phoenix KokoCarb Purewell K Series Katadyn Gravidyn Ecofiltro Orinko
Published PFAS analyses Published proofPFOA, PFOS and PFHxA monitored up to 3,200 L per filter Partial proofPFOA directly tested over 25 gal, about 95 L; the other compounds presented in the report rely on a surrogate approach No evidence identifiedPFAS/PFOA/PFOS not identified in the CF163W datasheet reviewed Published proofPFOA + PFOS: published combined reduction of 95%; cumulative test volume not stated Partial proofSeven PFAS tested at start-up only; no follow-up in litres Partial proofIAPMO report on seven PFAS; detailed tables up to 35 L, while the report states a 100 L test Partial proofPFOA and PFOS monitored up to 1,500 L No evidence identifiedThe standard K8012 compared here is not presented as a PFAS filter; no public PFAS report linked to this configuration was identified No evidence identified Partial proofIndependent report on three filters; one-off test on 1 L with heterogeneous results depending on the unit tested Published proofTwenty PFAS analysed by an independent laboratory; filtered test volume not publicly stated
Published TFA analyses Published proofTFA monitored up to 3,200 L per filter No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified No evidence identified
Published bacteria analyses Published proofBacteria monitored up to 3,000 L per filter Published evidence, but one-offMicrobiological challenge over 4 L after filter conditioning; no follow-up up to the ≈ 11,356 L claimed Published proofKlebsiella terrigena tested up to 3,000 L per CF163W filter Published proofE. coli / Typhoid / Klebsiella: >99.99% published; cumulative volume not stated Partial proofBacteria monitored up to about 300 L No dedicated Classic test identifiedThe Klebsiella test published by Euroguard concerns the carbon + UF Combi filter, not the Classic compared here Partial proofKlebsiella tested up to around 757 L Partial proofE. coli report over 5 L, dated 2020 and relating to a K8627 system, not the current K8626 Partial proofBachema tests on several bacteria cited by Katadyn; filtered volume not publicly stated Published evidence, but one-offE. coli and coliforms tested; no public follow-up over the claimed 2-year service life Published evidence, but one-offE. coli, coliforms and enterococci published; filtered test volume not stated
Published virus / viral marker analyses Published proofViruses monitored up to 3,000 L per filter Published evidence, but one-offMS2 and fr coliphages tested in a 4 L challenge; no follow-up up to the claimed service life Published proofRotavirus tested up to 3,000 L per CF163W filter Not claimedDoulton explicitly states that British Berkefeld elements do not filter viruses Partial proofMS2 bacteriophage tested at start-up only No evidence identified for the Classic Partial proofRotavirus tested up to about 757 L No evidence identifiedNo viral report clearly linked to the K8626 / K8012 identified No evidence identified No evidence identified No evidence identified
Overall reading of the level of proof Monitored up to 3,000 / 3,200 L per filterBroad contaminant spectrum; nominal service life of 6,000 L per pair (3,000 L per filter) Broad test file, often limited volumesNumerous independent reports, but a major gap between several tested volumes and the 3,000 gal claimed per element Numerous tests at 3,000 L per filterSubstantial public documentation; performance varies by contaminant and no PFAS/TFA identified Good documentation, but volumes often not statedStrong certifications and many published performance figures, but service life in litres is less clear Real tests, often at start-upBacteria up to 300 L; many other families tested only on a new filter Targeted tests up to 300 LReal IAPMO test file, but tested volumes far below the 3,000 gal claimed Real tests at intermediate volumesPFAS 1,500 L, NSF® chlorine 4,000 L; no consistent follow-up up to the 20,000 L claimed per pair Claimed capacity far above the tested volumes≈ 22,700 L claimed per pair; quantifiable public reports over 3 to 5 L and on an older K8627 Microbiology documented, volume not statedTechnically credible brand, but little public data on PFAS/TFA and on actual tested volumes Several recent tests, mostly one-offPFAS tested on 1 L; no public follow-up up to the 2-year service life Several recent tests, without follow-up to 11,350 LPFAS, bacteria and microplastics published; volumes often unstated or very low

The mentions “partial proof”, “claimed” or “no accessible proof” describe only the documentation identified. For Star Water Filter®, NSF® certificates for certain families of filter media are verified and held internally, but are not published so as not to disclose suppliers and the filter formulation; they are therefore not presented here as public evidence. The public NSF® certification of Ultimate Star Filter® USF-SWF-001 is shown with its exact scope. The most useful approach is to compare what is tested, over how many litres, under which protocol and how closely this approaches the claimed service life.

BRAND-BY-BRAND READING

What the table actually shows, brand by brand

For each brand, the review below first compares what is claimed, what has actually been tested and, when the information is available, over how many litres. This is often more meaningful than a certification logo or a percentage obtained on a new filter. Flow rate is also taken into account because a filter must remain practical in daily use. For a broader explanation, see also how to tell whether a water filter is truly effective.

Star Water Filter®

Most comprehensive documentation

Quick read

Designed in France, Ultimate Star Filter® has a claimed service life of 6,000 L per pair, i.e. 3,000 L per filter. Chemical analyses are published up to 3,200 L per filter and microbiology up to 3,000 L per filter, by independent ISO/IEC 17025-accredited laboratories. Readers can therefore directly compare the claimed service life with the volumes actually tested.

Key points

  • French design and documented multi-layer architecture.
  • Independent ISO/IEC 17025-accredited laboratories and full reports published.
  • Chemical analyses up to 3,200 L per filter, beyond the nominal service life of 3,000 L per filter.
  • Microbiology up to 3,000 L per filter: bacteria, Salmonella and viruses.
  • Broad spectrum analysed through end of service life: PFAS and TFA, heavy metals, pharmaceutical residues, plastic derivatives, hormones, pesticides, VOCs, halogenated VOCs and microplastics.
  • One of the highest flow rates in the comparison: 12 to 16 L/h with a new pair, then around 4 L/h at about 3,000 L.
  • Ultimate Star Filter® NSF® listed; supplier certifications for certain media are verified and held by SWF.
  • Published REACH compliance: European compliance regarding chemical substances present in materials.

Good to know

Flow rate naturally decreases with use: it typically moves from 12–16 L/h with a new pair to around 4 L/h at about 3,000 L. This evolution is published, so users know what to expect over the filter’s service life.

Why SWF stands out

Ultimate Star Filter® combines several advantages that are rarely found together: French design, independent analyses up to 3,000–3,200 L per filter across a broad and coherent range of pollutants, fast yet controlled flow rate, NSF® certification and REACH compliance. With a nominal service life of 6,000 L per pair (3,000 L per filter), the main advantage is simple: performance is documented through the filter’s service life, and even beyond it for most chemical analyses.

Berkey / Black Berkey®

Quick read

Berkey has built strong recognition over many years. However, a pair of Black Berkey® elements is claimed to last about 22,700 L, while the public tests identified cover much lower volumes depending on the contaminant. The available results therefore do not demonstrate that the same performance is maintained through the claimed service life.

What the data show

  • Around 22,700 L claimed for a pair of Black Berkey® elements.
  • Flow rate up to 13.3 L/h on the 8.5 L Big Berkey® with two elements and a full upper chamber.
  • PFOA directly tested up to about 95 L in the report identified.
  • Microbiological challenge identified over about 4 L of contaminated water after filter conditioning.
  • Several of the technical reference reports identified are old, particularly those concerning metals and PFAS.

What to put in perspective

A good result over 4 L, 95 L or a few hundred litres does not show that the same performance will still be present after more than 22,000 L for the pair. Black Berkey® is not the filter used in the Berkey systems currently NSF® listed with Phoenix elements: that is a different configuration.

Overall reading

The key point with Black Berkey® is the gap between a very high claimed capacity and the volumes actually covered by public analyses. The brand’s long-standing reputation does not replace verification of performance after several thousand litres.

Coldstream FTO Plus (CF163W)

Quick read

Coldstream publishes results for a substantial number of contaminants, including several measurements carried out up to 3,000 L per CF163W filter. The published nominal capacity is 1,500 L per filter. However, no PFAS or TFA analyses were identified in the datasheet reviewed, and some performance figures decline significantly with volume.

What the data show

  • IAPMO R&T tests according to NSF/ANSI 42, 53 and P231; laboratory compliant with ISO 17025.
  • Klebsiella, Rotavirus, numerous metals, VOCs, pesticides, pharmaceutical residues and hormones measured up to 3,000 L per CF163W filter.
  • Claimed nominal capacity: 1,500 L.
  • Published minimum flow rate: 1 L/h per CF163W filter.

What to put in perspective

  • At 3,000 L per filter, some reductions fall sharply: free chlorine is published at 80.5% and fluoride at 17.05%.
  • No PFAS or TFA analysis identified in the CF163W datasheet reviewed.
  • The tests are conducted according to several NSF/ANSI standards, but this should not be confused with an NSF® listing of the CF163W if no such listing is identified.

Overall reading

Coldstream documents several performance figures up to 3,000 L per CF163W filter, beyond its nominal capacity of 1,500 L per filter. Contaminants must nevertheless be considered one by one: some performance figures remain high, while others fall sharply. The published minimum flow rate is also 1 L/h per filter.

British Berkefeld / Ultra Sterasyl

Quick read

British Berkefeld publishes technical data and certifications across several criteria. The main limitation when comparing performance over time is that the actual tested volume is not always stated depending on the contaminant. Its ceramic technology also results in a slow flow rate: around 1 L/h per cartridge.

What the data show

  • Sterasyl® ceramic + granular activated carbon + lead-reduction media.
  • Published NSF/ANSI certifications on several criteria, including particulates, cysts/turbidity and microplastics depending on the model.
  • Bacteria, Cryptosporidium and Giardia: published performance above 99.99% in Doulton tables.
  • PFOA + PFOS: 95% combined reduction published; cumulative test volume not stated on the data page reviewed.
  • Typical flow rate: 1 L/h per Ultra Sterasyl, i.e. around 2 L/h with two cartridges.

What to put in perspective

  • Replacement recommended after 6 months, with no nominal capacity in litres published for Ultra Sterasyl.
  • The cumulative test volume is not always stated, making it impossible to know how far performance has actually been verified.
  • Doulton states that British Berkefeld elements do not filter viruses.
  • The slow flow rate of the ceramic can become restrictive in daily use.

Overall reading

British Berkefeld provides useful data on several contaminants, but performance over time remains difficult to assess when volumes are not published. Another very practical difference compared with SWF is its much slower filtration flow rate.

Weeplow O’Pure 2

Quick read

The O’Pure 2 pair is claimed to last 22,000 L and a flow rate of 3.8 L/h. Bacteria are tested up to about 300 L, while PFAS, MS2 virus, pesticides and metals are mainly tested on a new filter or at start-up.

What the data show

  • Activated carbon + hollow-fibre ultrafiltration 0.02 µm.
  • Bacteria monitored up to about 300 L.
  • Seven PFAS analysed at filter start-up.
  • MS2 virus, pesticides and heavy metals tested on a new filter / at start-up according to the reports identified.

What to put in perspective

The difference between 22,000 L claimed and tests at 300 L or at start-up remains considerable. These reports show performance under the tested conditions, but not that it is maintained after several thousand litres.

Overall reading

The key point with Weeplow is not the number of reports displayed, but the gap between the claimed service life and the actual tested volume for each contaminant family.

Euroguard Classic

Quick read

Euroguard Classic is claimed to last 3,000 gallons per cartridge, or about 11,356 L. However, the public analyses identified cover much lower volumes: a few dozen to a few hundred litres depending on the contaminant.

What the data show

  • Chlorine tested up to 300 L.
  • Lead, cadmium and mercury tested up to 50 L.
  • Seven PFAS analysed; visible result tables up to 35 L in the report reviewed.
  • No clear numerical flow rate identified for the two-cartridge Classic configuration on the public pages reviewed.

What to put in perspective

  • The published Klebsiella bacterial test concerns the carbon + UF Combi filter, not the Classic compared here.
  • The NSF® listing identified covers material / structural-integrity requirements; it does not prove PFAS reduction over 11,356 L.
  • The PFAS report states 100 L in its summary, while the visible result tables go up to 35 L.

Overall reading

Compared with the roughly 11,356 L claimed per cartridge, tests at 35, 50 or 300 L remain far too distant from the claimed end of life to demonstrate sustained performance over time.

Phoenix KokoCarb

Quick read

Phoenix claims 20,000 L for a pair of KokoCarb cartridges and a flow rate of up to 16 L/h. However, the actual published volumes vary widely by contaminant: about 757 L for microbiology, 1,500 L for PFOA/PFOS and 4,000 L for chlorine.

What the data show

  • Silver-impregnated coconut-shell activated-carbon block.
  • PFOA/PFOS up to 1,500 L in the report identified.
  • NSF/ANSI 42: chlorine / taste / odour reduction up to 4,000 L for PCC8/KokoCarb.
  • Klebsiella and Rotavirus tested up to about 757 L in the reports identified.
  • Claimed flow rate: up to 16 L/h with two cartridges.

What to put in perspective

The issue is not the absence of tests, but their varying depth: 757 L, 1,500 L or 4,000 L depending on the contaminant, for a pair claimed to last 20,000 L. Phoenix also recommends replacing the cartridges every 12 months. No TFA analysis was identified.

Overall reading

Phoenix combines a high flow rate with several published tests, but the data do not show that all performance remains unchanged up to the claimed 20,000 L.

Purewell K Series Classic (K8626 / K8012)

Quick read

The Purewell K Series Classic K8626 uses two K8012 filters and claims up to 6,000 gallons per pair, about 22,700 L, with a flow rate of up to about 15.1 L/h. However, the public performance reports identified are dated 2020, concern an older K8627 system and cover only 3 to 5 L for the quantifiable tests.

What the data show

  • K8012 architecture: activated carbon + silver ions + 0.01 µm UF membrane.
  • Current advertised capacity: up to 6,000 gallons per pair, about 22,700 L.
  • Claimed flow rate: up to 4 gal/h with two filters, i.e. about 15.1 L/h.
  • Purewell provides a public centre gathering certificates and performance reports.

What to put in perspective

  • Bacterial report identified: 5 L, on a 2020 K8627 system.
  • Chlorine / lead / mercury report identified: about 3 L, also on a 2020 K8627.
  • These reports do not concern the current K8626 system, although they relate to a previous-generation Purewell gravity system.
  • No identified public follow-up makes it possible to verify performance up to the roughly 22,700 L claimed for the pair.
  • No public PFAS report was identified for the standard K8012 filter compared here.

Overall reading

The current K Series advertises a high capacity and flow rate, but the comparison should focus above all on what has actually been tested: the quantifiable public reports found cover only a few litres and an older K8627 model, far from the currently claimed capacity.

Katadyn Gravidyn

Quick read

Katadyn Gravidyn combines silver-treated microporous ceramic and activated carbon, with replacement recommended no later than after 6 months. Katadyn cites independent microbiological tests, but the public documentation reviewed does not state how many litres had been filtered during these tests.

What the data show

  • Bachema tests cited on several bacteria, including E. coli and Salmonella.
  • Ceramic with incorporated silver + high-performance activated carbon.
  • Published Gravidyn flow rate: 4 L/h.
  • Replacement recommended after 6 months.

What to put in perspective

  • Microbiological test volume not stated in the public documentation identified: it is impossible to know whether performance was verified after hundreds or thousands of litres.
  • No public file identified on PFAS, TFA or viruses for Gravidyn.
  • The 4,000 L figure sometimes associated with Katadyn concerns another product and is not used for Gravidyn.

Overall reading

Katadyn publishes microbiological information, but without a public test volume these results cannot be turned into evidence of performance over time. This missing information limits comparison with filters tested over several thousand litres.

Ecofiltro

Quick read

Ecofiltro recommends replacing its filter unit after 2 years, with no clearly published nominal volume in litres. The 20 L model filters about 1 to 2 L/h. Several recent reports exist, but the tests identified remain one-off.

What the data show

  • Architecture: clay/ceramic + sawdust-derived activated carbon + colloidal silver.
  • Public reports on bacteria, PFAS, microplastics, pesticides, heavy metals and pharmaceutical residues.
  • Ecofiltro 20 L flow rate: 1 to 2 L/h.

What to put in perspective

  • The Measurlabs PFAS report covers 1 L per unit.
  • The three units tested give heterogeneous results for some PFAS.
  • No identified public follow-up links these one-off tests to the claimed service life of 2 years.
  • No public TFA or viral test identified.

Overall reading

Ecofiltro now has several reports, but the most important information for service life is still missing: up to how many litres have these performance figures been verified?

Orinko

Quick read

Orinko claims 11,350 L per cartridge, i.e. 22,700 L with two cartridges, and a flow rate of 8 L/h with two cartridges. Analyses are published, but the filtered volume is not stated for several important tests.

What the data show

  • 20 PFAS analysed; cumulative filtered volume not stated in the public information identified.
  • Microplastics tested by FILAB over 1 L.
  • Bacteria analysed with before/after results; cumulative volume not clearly stated publicly.
  • Claimed flow rate: 4 L/h per cartridge, i.e. 8 L/h with two.
  • NSF® certification must be read according to the exact scope of the official listing, separately from contaminant-performance reports.

What to put in perspective

A before/after analysis without a stated cumulative volume shows what happens under the test conditions, but not after several thousand litres. For microplastics, the volume is known: 1 L tested, compared with the 11,350 L claimed per cartridge.

Overall reading

With Orinko, the key point is therefore simple: 11,350 L claimed per cartridge, but several analyses do not state the tested volume, and the microplastics test was performed on 1 L. This does not document performance through the claimed service life.

SCIENTIFIC CRITERIA

The most useful criteria for seriously comparing a gravity water filter

Not all criteria carry the same weight in a comparison. Some provide surface information, while others allow a finer reading of the technical consistency, the quality of the evidence and real credibility of a system.

01

Filter architecture

The internal composition of a filter largely determines how it works. Activated carbons, ceramic, hollow fibres or a nanofiltration membrane do not serve the same purposes or the same families of contaminants. In gravity filtration, this architecture must also be read alongside the flow rate and the contact time between water and filter materials.

A system whose architecture is clearly documented is generally easier to compare than a system whose components remain vague or are merely mentioned.

02

Presence of a nanofiltration membrane

The presence of a nanofiltration membrane significantly changes the technical profile of a gravity system. It can strengthen retention capacity on certain complex contaminants, pathogens or residues that are difficult to address with activated carbon alone.

It is therefore a structuring criterion, provided it is genuinely documented and consistent with the published analyses.

03

Long-term stability

An announced service life does not, by itself, provide sufficient information. The decisive point is whether performance remains readable across the litres filtered and up to volumes close to the recommended end-of-life.

The documented long-term stability is often the most useful criterion for distinguishing a commercial promise from performance that is actually monitored.

04

Level of proof and protocol

Manufacturer data, certification, regulatory compliance and independent analysis do not provide the same level of information. Comparing them without distinction often leads to hasty conclusions.

A serious comparison must rank the evidence and prioritise what can be verified, reread and placed within an identifiable protocol, such as laboratory analyses. This is especially true when test conditions strongly influence results, for example for certain heavy-metals tests.

05

Demanding contaminants and test conditions

Certain emerging or highly discussed contaminants, such as PFAS, the TFA or certain heavy metals, require more rigorous reading than standard tests. Their presence in a comparison can reveal a higher level of requirement in the published documentation.

It is still necessary to look at the volume at which these analyses are performed, and under which conditions. For heavy metals, tests conducted under conditions comparable to NSF® 53, especially at pH 6.5 and 8.5, provide a much stronger reading.

06

Material compliance, REACH and system consistency

The quality of a system depends not only on what it filters, but also on what it puts in contact with water. The material compliance, when clearly published, provides useful information on the manufacturer’s transparency.

This consistency can also be seen in the system’s complementary technical choices, for example the presence of a silver-based bacteriostatic treatment. This type of element does not replace filtration analyses, but it complements the reading of a system when architecture, materials and performance are documented together.

07

Filtration flow rate and ease of use

A filter may look effective on paper but become restrictive if users have to wait hours for a few litres of water. The flow rate is therefore a genuine practical criterion, provided comparable configurations are used and the figure is understood in context: full chamber, new filter or minimum value.

A fast flow rate does not, on its own, prove good filtration. The real value lies in combining practical flow rate with performance verified over time. With Ultimate Star Filter®, a new pair typically filters 12 to 16 L/h, with around 4 L/h still available at about 3,000 L of use, while performance is analysed up to 3,000–3,200 L per filter.

Taken in isolation, no single criterion is sufficient. The most useful reading is based on their overall consistency: architecture, level of proof, stability over time, depth of analyses and test conditions.

CRITICAL READING

Common mistakes when comparing gravity water filters

Some indicators are often interpreted too quickly. These shortcuts can create a misleading impression of performance or safety. A more rigorous reading helps distinguish more clearly what belongs to a promise and what is based on genuinely documented proof, especially for a gravity water system.

01

A high service life is not enough to judge a filter

An announced service life (expressed in litres or months) does not guarantee that performance remains stable up to that threshold. Without intermediate or end-of-life data, this information remains partial.

See also: service life of a gravity water filter.

03

One test is not enough to prove performance

An isolated result, carried out at low volume or using a poorly detailed protocol, does not allow performance to be read over time. Reproducibility, the progression of tested volumes and test conditions are essential.

For certain contaminants such as heavy metals, analysis conditions directly influence results. Tests performed under conditions comparable to NSF® 53, especially at pH 6.5 and 8.5, provide a much more robust reading.

See also: laboratory test results.

04

Results on a few contaminants are insufficient

Testing a limited number of substances does not allow conclusions to be drawn about the overall capacity of a system. The diversity of contaminants analysed and their level of requirement provide a more reliable reading.

This is particularly true for PFAS, the TFA, heavy metals or certain emerging contaminants.

05

A technical component is not enough to reassure

The presence of a membrane, a specific carbon or a bacteriostatic treatment (such as silver) does not, on its own, constitute proof of performance. These elements must fit into a consistent architecture and be confirmed by published analyses.

See also: design & architecture.

06

Detailed communication is not proof

A technical or marketing description, even a precise one, does not replace independent validation. The difference between declared information and measured proof is essential.

This is why it is useful to distinguish manufacturer data from actually published analyses and certifications.

07

A slow filter is not automatically more effective

Slowness is not proof of quality. A filter can have a low flow rate and remain poorly documented, while a faster filter can maintain very good performance if its architecture and media are suitable.

The right question is therefore: is the claimed flow rate compatible with analyses carried out after several hundred or several thousand litres?

A useful comparison is less about accumulating indicators than about checking their consistency, their level of proof and the conditions under which they are obtained, in relation toarchitecture, the stability over time and the published analyses.

DECISION SUPPORT

Which gravity water filter should you choose according to your level of requirement?

Choosing a system depends less on a ranking than on how you read the criteria. Depending on your level of requirement in terms of proof, stability and documentation, some options appear more consistent than others. To go further, see also the buying guide for your gravity water filter.

Daily use and simple approach

For standard domestic use, several systems may be suitable if the objective is to improve the taste of water or reduce certain standard contaminants.

In this case, more accessible solutions may be considered, provided that you accept a more limited reading of long-term performance and level of proof.

Demanding contaminants and advanced reading

When specific concerns are present (PFAS, TFA), pathogens, complex residues), the comparison must go beyond standard tests.

Systems capable of documenting these dimensions with in-depth and time-consistent analyses provide a more robust and reassuring reading.

High promises but limited documentation

Some systems display long service lives or extensive performance claims, without publishing data that allows their real stability to be verified.

In this case, comparison must remain cautious, clearly distinguishing what is announced from what is actually documented.

As of the date of writing, the most relevant systems are those that show overall consistency between architecture, published analyses, stability over time and accessible level of proof.

KEY TAKEAWAY

What truly distinguishes gravity water filters

  • Readable architecture → clearly identified and consistent components
  • High level of proof → published independent analyses
  • Credible test conditions → protocols comparable to recognised standards (e.g. NSF® 53)
  • Depth of analyses → PFAS, TFA, heavy metals, bacteria, viruses
  • Long-term stability → results followed up to volumes close to end-of-life
  • Practical flow rate → sufficient filtration speed for daily use, without sacrificing evidence of performance

Essential reading

A high-performing filter is not recognised by a single indicator, but by the consistency between design, analyses, tested volume, flow rate and test conditions.

The most accomplished systems are those that make these elements readable, verifiable and consistent, including in particular the number of litres over which performance was actually tested.

To go deeper into your choice, see the complete buying guide or discover the criteria that distinguish the most accomplished systems.

TO GO FURTHER

Choose a documented filter, not just an announced filter

Among the 11 systems reviewed, Star Water Filter® presents the most comprehensive documentation according to the criteria selected. Ultimate Star Filter® combines French design, analyses carried out by accredited independent laboratories, a broad spectrum of pollutants tested up to 3,000 to 3,200 L per filter, a fast yet controlled flow rate, NSF® certification and REACH compliance. Its nominal service life is 6,000 L per pair, i.e. 3,000 L per filter: the analysis volumes are therefore directly comparable with the claimed service life.

Discover Star Water Filter® filters

SOURCES

Sources and documents reviewed

The data in this comparison were rechecked using documents accessible as of 20 August 2026. We prioritise laboratory reports, official certification registers and manufacturers’ technical pages. When a report does not state the filtered volume, we say so explicitly.

Manufacturers’ websites and documentation may change. The links below are the main sources used for the current version of this comparison.

For the flow rate, we use, wherever possible, a standard configuration with two filter elements, a very common setup among the gravity systems compared. When a manufacturer publishes only a value per cartridge or per system, we keep that figure as published and state it clearly. When no clear numerical value is published, we simply state “no public numerical value”.

Star Water Filter®

Berkey / Black Berkey®

Coldstream FTO Plus

British Berkefeld / Ultra Sterasyl

Weeplow O’Pure 2

Euroguard Classic

Phoenix KokoCarb

Purewell K Series

Katadyn Gravidyn

Ecofiltro

Orinko

SWF note: some NSF® certifications for media used in the Ultimate Star Filter® formulation are held internally and intentionally not published, because disclosure could identify suppliers and formulation components. They are therefore kept separate from the public documents listed above.

Analysis methodology

Methodology

Analysis based on public data, laboratory test protocols and a critical reading of long-term performance.

Author Star Water Filter® technical team

This content is written by the Star Water Filter® technical team, specialising in the analysis of gravity-filtration systems. It is based on public documents, independent laboratory reports, certification registers and a comparison of actual tested volumes, published flow rates and claimed service lives. When SWF information is documented internally but confidential, this distinction is explicitly stated.

FAQ

Frequently asked questions about the gravity water filter comparison

This FAQ complements the comparison by answering the most frequent questions about comparison criteria, proof reading and the choice of a gravity filtration.

What is the best gravity water filter?

There is no single best gravity water filter in absolute terms: the right choice depends in particular on intended use, budget, required capacity and the level of evidence expected. Among the 11 systems reviewed here, Star Water Filter® nevertheless presents the most comprehensive documentation according to the criteria selected: nominal service life of 6,000 L per pair (3,000 L per filter), published chemical analyses up to 3,200 L per filter, microbiology up to 3,000 L per filter, PFAS/TFA and detailed architecture. This conclusion relates to the level of documentation and evidence assessed, and does not constitute a universal ranking across every possible criterion.

How can gravity water filters be compared seriously?

A serious comparison is not just about comparing brands or marketing promises. Manufacturer data, certifications, regulatory compliance and independent analyses must be distinguished. The most useful criteria are generally the filter architecture, the documented long-term stability, the actual tested volume, the depth of published analyses and the filtration flow rate for daily use.

Why is the announced service life not sufficient?

An announced service life in litres or months remains declarative information if it is not accompanied by analyses performed at different stages, especially near the recommended end-of-life. The real issue is whether performance remains readable over time, and not only at the beginning of use.

Does NSF® certification always guarantee filtration performance?

No. An NSF® mention must always be read in detail, because it can cover different aspects depending on the case. Some certifications concern materials or sanitary safety, while others directly concern filtration. It is therefore important to check precisely what the certification covers. See also: NSF® certification and the Ultimate Star Filter® water filter.

Why is long-term stability a central criterion?

Because a new filter does not tell the whole story about how the system behaves after several hundred or thousand litres. Documented stability makes it possible to read whether performance remains consistent over use. It is often the most useful criterion for distinguishing an announced promise from actually monitored performance.

Why is filter architecture important?

Architecture largely determines the technical profile of the system. Activated carbons, ceramic, hollow fibres or a nanofiltration membrane do not play the same role. Understanding a filter’s architecture helps better interpret published analyses and check whether results are consistent with the components used.

Should PFAS and TFA be looked at in a comparison?

Yes, when the documentation includes them. The PFAS, the TFA and certain demanding contaminants often help distinguish the most rigorous systems in their analytical approach. But these results must always be read with their protocol, test volume and place within the overall proof system.

What should be remembered from this comparison when choosing a system?

The most important point is to compare consistent evidence rather than isolated promises. As of the date of writing, the most convincing systems are those that make their architecture readable, publish independent analyses, document long-term stability and provide a clear level of proof on the truly decisive criteria. For more practical guidance, you can also consult the buying guide.