In real homes, properly installed filters certified for lead under NSF/ANSI 53 brought lead below 1–5 µg/L. In Flint, over 97% of samples were below 0.5 µg/L. Failures come from very fine lead particles, very high incoming lead and wrong installation. Iron in the water can clog cartridges early. We found no trial that measured children's blood lead with and without a filter.
| Source | What it found (quoted) and its limits | Evidence |
|---|---|---|
| Tully J, Schock M, Shilling S, Bosscher V, Lytle D, Harmon S, Bennett-Stamper C. An evaluation of properly operated NSF/ANSI-53 Pb certified drinking water filters in Benton Harbor, MI. J Water Health 22(2):296-308 (2024) · new peer-reviewed field study · Benton Harbor MI, 9 Nov-17 Dec 2021 · n = 199 filters |
filtered Pb <5 ppb for all; maximum 2.5 ppb; discrete Pb nanoparticles (<100 nm) not widely found; some residents confused about installation/use (abstract only) Limits: abstract only; same dataset as A05; low influent lead |
Evidence: strong Free to read · source · added 2026-10-30 |
| US EPA Region 5. Data Report: Summary of Lead Water Results in Filter and Sequential Studies, Benton Harbor MI (2022) · new government field test report · Benton Harbor MI homes · n = 215 homes sampled; 199 properly installed/operated filters |
unfiltered <0.5 to max 77 ppb; filtered: 95% (291 samples) <1 ppb; 90% (277) <0.5 ppb; 100% <5 ppb; max 2.5 ppb Limits: Report inconsistent on faucet/pitcher split (89/11 vs 72/28 %); 79% of unfiltered samples already <5 ppb (low challenge) |
Evidence: strong Free to read · source · added 2026-10-30 |
| Bosscher V, Lytle DA, Schock MR, Porter A, Del Toral M. POU water filters effectively reduce lead in drinking water: a demonstration field study in Flint, Michigan. J Environ Sci Health A 54(5):484-493 (2019) · new field study (US EPA) · Flint MI homes, 2016 · n = over 345 locations |
'Over 97% of filtered water samples contained lead below 0.5 µg/L'; 'maximum lead concentration in filtered water was 2.9 µg/L' (EPA summary page; journal abstract not readable - tandfonline blocked) Limits: Numbers taken from EPA summary page, not full text; filters properly installed; influent max not given on page |
Evidence: strong Free to read · source · added 2026-10-30 |
| Purchase JM, Villalona C, Katner A, Pieper K, Edwards M. Household Point-of-Use Faucet Filters for Lead Removal: Field Performance and User Experiences. ACS ES&T Water 5(6):3045-3053 (2025) · new field study + lab · New Orleans and Enterprise, Louisiana · n = 21 occupied homes + 2 unoccupied high-risk homes |
occupied homes: all filtered <1 µg/L; ELA avg removal 94.8%; disturbed lead service line home: influent particulate 9-3053 µg/L, 8 of 20 samples >10 µg/L before 100% capacity, removal 21-99%; ELA filters clogged at avg 62% flow loss, most replaced before 50% rated capacity; 10% (n=88) filtered samples >50 µg/L Mn Limits: Small n; one brand; clogging judged by residents |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Purchase JM, Rouillier R, Pieper KJ, Edwards MA. Understanding Failure Modes of NSF/ANSI 53 Lead-Certified Point-of-Use Pitcher and Faucet Filters. Environ Sci Technol Lett 8(2) (2021) · new lab challenge study · lab, extreme conditions · n = 10 brands |
tested at '<= 200% of rated capacity' and influent ~1000 µg/L; most brands effluent <10 µg/L; with colloidal lead some replicates passed 50% of influent lead; some exceeded 5, 10 and 15 µg/L; failures: leaks, premature clogging, hole in medium Limits: abstract only; extreme challenge, not real tap water |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Doré E, Formal C, Muhlen C, Williams D, Harmon S, Pham M, Triantafyllidou S, Lytle DA. Effectiveness of point-of-use and pitcher filters at removing lead phosphate nanoparticles from drinking water. Water Res 201:117285 (2021) · new lab study, synthetic water · US EPA lab · n = 6 filters (3 faucet, 3 pitcher), triplicate |
challenge ~100 µg/L Pb, particles 0.016-0.098 µm; POU effluent mean 34.3-56.4 µg/L (44.6-65.1% reduction); Pitcher-1 83.1 µg/L (~19%), Pitcher-2 88.2 µg/L (~11%), Pitcher-3 7.5 µg/L (~93%); only Pitcher-3 <10 µg/L, none <5 µg/L Limits: Synthetic low-hardness water, authors call it likely extreme case; filters used to 4-35% of capacity; not NSF protocol |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Mulhern R, MacDonald Gibson J. Under-Sink Activated Carbon Water Filters Effectively Remove Lead from Private Well Water for over Six Months. Water 12(12):3584 (2020) · new field longitudinal study · private wells, 2 North Carolina counties, Oct 2019-Jun 2020 · n = 17 homes |
mean removal 98%; all effluent <1 µg/L; 95% of effluent <0.5 µg/L; influent beneath sink mean 3.3 µg/L (max 8.4); 3 of 17 filters (~18%) clogged after 2-3 months Limits: Low influent lead; small n; 10% paired samples excluded; 2 months lost (COVID) |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Pan W, Johnson ER, Giammar DE. Accumulation on and extraction of lead from point-of-use filters for evaluating lead exposure from drinking water. Environ Sci: Water Res Technol 6:2734-2741 (2020) · new lab study, real tap water · St. Louis tap water spiked · n = not stated in summary (replicate filters) |
removed >99% (avg) of dissolved Pb from tap water spiked to 50 µg/L over 100 gal (379 L); effluent near/below 0.5 µg/L detection limit; last 10 gal still <10 µg/L Limits: Dissolved lead only; one water; primary aim is sampling method |
Evidence: moderate Free to read · source · added 2026-10-30 |
| City of Newark Dept of Water and Sewer Utilities. Point-of-Use Filter Study: Filter Results Report (Final), Aug-Sept 2019 (2019) · new government/utility field study (with CDM Smith, NJDEP, EPA) · Newark NJ homes with lead service lines · n = 198 properly installed filters (189 faucet, 9 pitcher); 591 filtered samples |
unfiltered max: first draw 306.0 ppb, 5-min flush 392.0 ppb; 97.5% (193/198) of filters had all samples <=10 ppb; 98.8% (584/591) filtered samples <=10 ppb; 91.9% filters all samples <=5 ppb; failure case filtered first draw 32.6 ppb (unfiltered 306 ppb, above 150 ppb NSF challenge) Limits: Utility self-report; 67 of 265 filters excluded as not properly installed/maintained; last ~16k chars of PDF not read |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Brown KW, Gessesse B, Butler LJ, MacIntosh DL. Potential Effectiveness of Point-of-Use Filtration to Address Risks to Drinking Water in the United States. Environ Health Insights 11 (2017) · new narrative review + secondary analysis of Flint EPA data · US; Flint MI · n = 208 matched sample pairs (Flint) |
Flint: median unfiltered Pb 2.5 µg/L vs 0.11 µg/L filtered (>20-fold); highest filtered 2.9 µg/L when unfiltered ~419 µg/L (Table 1: 418); 90th percentile filtered 0.5 µg/L; literature: SBAC removed 80% to 99% of total Pb in 2 studies Limits: Authors affiliated with consulting firm (check funding); not systematic; reuses EPA Flint data (overlap with A01/A02) |
Evidence: moderate Free to read · source · added 2026-10-30 |
| US EPA / ATSDR. Flint Filter Challenge Assessment Summary (field report) (2016) · new government field test report · Flint MI taps · n = over 200 taps + at least 50 additional locations |
filtered average 'just under 0.3 ug/L' (non-detects at detection limit); about 80% of results below detection; <1% of MDEQ residential samples exceeded 150 ug/L Limits: No max filtered value or % <1 µg/L reported; overlaps with A01 dataset |
Evidence: moderate Free to read · source · added 2026-10-30 |
| Aljadani A. Efficacy of Pitcher Filters For Drinking Water Lead Reduction: A Challenge and Demonstration Study in Denver, CO. MS thesis, Univ. Colorado Boulder (2023) · new lab challenge + field (thesis) · Denver Water corrosion racks + field homes 2020-2021 · n = 5 pitcher filters; at least 50 field sites |
challenge Pb 31-97 ppb, all 5 filters removed >94%; 2 filters (E, C) not consistently <5 ppb; field: all below action level except one sample where filtered 40.8 ppb > unfiltered 1.4 ppb Limits: abstract only; thesis; anomalous field sample unexplained |
Evidence: weak Free to read · source · added 2026-10-30 |
| Lytle DA, Formal C, Doré E, et al. Synthesis and characterization of stable lead (II) orthophosphate nanoparticle suspensions. J Environ Sci Health A (2020) · new lab method study · US EPA lab · n = n/a |
stock 5 mg/L Pb; challenge 0.10 mg/L Pb; particles avg 38 nm; motivated by Pb nanoparticles passing certified filters in Newark NJ (abstract) Limits: abstract only; no filter performance data itself |
Evidence: weak Free to read · source · added 2026-10-30 |
| Rouillier RJ. Understanding Practical Limitations of Lead Certified Point of Use (POU) Filters. MS thesis, Virginia Tech (2020) · new lab study (thesis) · lab, up to 4 waters · n = 10 brands |
filters usually reduced lead by at least 80%; sometimes >5 µg/L and occasionally >15 µg/L with particulates; sometimes <50% particulate Pb removed; clogging '50% (3/6) ... at an iron concentration of 0.37 mg/L', 66% (4/6) at 1 mg/L, 100% (6/6) at 20 mg/L Limits: abstract only; thesis (not peer reviewed); overlaps with A09 |
Evidence: weak Free to read · source · added 2026-10-30 |
| Deshommes E, Zhang Y, Gendron K, Sauvé S, Edwards M, Nour S, Prévost M. Lead removal from tap water using POU devices. J AWWA 102(10):91-105 (2010) · new lab/field with tap water · Montreal (Canada) · n = unknown |
SECONDARY (as cited by Pan et al. 2020, A12): 80-92% removal of dissolved lead and 92-99% removal of particulate lead Limits: abstract not retrieved; numbers secondhand - verify before use |
Evidence: weak Paywalled · source · added 2026-10-30 |
Evidence grade: strong = trials, large field studies or official datasets; moderate = smaller or lab studies with real water; weak = single small studies, makers' specs or secondary reports. "Free to read" means the source itself is free to read online.