The win is decades, not days — a small but real cut in your lifetime odds of bladder and colorectal cancer (the disinfection-by-product story), bypassing lead from old plumbing, and ducking the per- and polyfluoroalkyl substances that don't leave your body. A jug-style pitcher fixes taste and chlorine and not much else; a certified carbon block at the sink fixes lead and most disinfection by-products too; a reverse-osmosis unit handles essentially everything at the cost of stripping out minerals (closeable with a $30 cartridge or a regular diet). The real failure mode isn't which type you bought — it's forgetting to change the cartridge.
Three completely different jobs are happening under the hood, and the differences explain why a $25 pitcher and a $700 under-sink unit aren't doing the same thing in smaller and bigger versions.
Activated carbon traps molecules on its surface. A teaspoon of activated carbon has the internal surface area of a football field — chlorine, disinfection by-products, many pesticides, and longer-chain PFAS stick to the pore walls and don't come out the spout. The catch: those sites fill up. Once the carbon is saturated, the filter stops removing — and in some cases starts releasing — what it had been trapping. Slow water flow and a fat bed of carbon are what make this work; a thin pitcher cartridge with a hard-poured glass is the worst case.
Ion exchange swaps one ion for another. Most pitcher cartridges blend the carbon with a resin that grabs heavy-metal ions out of solution and releases harmless ones back. This is how the better pitcher filters get any traction on lead at all. Like carbon, the resin has a finite capacity.
Reverse osmosis pushes water through a membrane with pores the size of a water molecule. Almost nothing else fits. Dissolved metals, organics, salts, microplastics — they get rejected to a drain line and only the water passes. This is why RO is the only home approach that handles short-chain PFAS, nitrate, and microplastics consistently, and also why it strips out calcium and magnesium along with everything else (more on that under contraindications).
One footnote worth knowing: about a fifth of US municipal supplies use chloramine instead of chlorine because it lasts longer in the pipes. Standard activated carbon barely touches chloramine — you need catalytic carbon (carbon treated to chemically break chloramine apart) or RO. Your utility's annual water-quality report tells you which disinfectant they use.
What the data actually shows each type removes
The marketing on every filter box is a Venn diagram problem: same words, very different overlaps. Strip the language back to what's been independently tested and the picture clarifies fast.
Chlorine and taste. Solved by any activated-carbon filter rated to NSF/ANSI 42. Pitcher, faucet-mount, under-sink, RO — all handle chlorine taste and odour. This is the single thing pitcher filters do reliably.
Lead. Only filters certified to NSF/ANSI 53 specifically for lead are reliable here. The standard requires a filter to knock a 150 µg/L challenge down to 5 µg/L or less — over 99% removal — at two different water chemistries. A review of 1,678 certified filters found 96% met benchmark in testing, and field-deployed certified filters in lead-affected homes hit 99% success (Tang et al. 2023). The catch is which filters carry the certification: the standard Brita pitcher does not; the Brita Elite cartridge does. ZeroWater pitchers do. Most faucet-mount and under-sink carbon blocks do. Refrigerator filters mostly don't. Check the box.
Disinfection by-products (DBPs). When chlorine reacts with leaves and other organic matter in source water it produces THMs and related compounds. The EPA caps total THMs at 80 parts per billion as a running annual average across the distribution system (EPA Stage 1 D/DBP Rule, 1998). Activated carbon at adequate contact time pulls these down well below the limit; RO does the same.
PFAS — the "forever chemicals." A Duke / NC State team tested 76 home filters in real households against PFOA, PFOS, and a handful of related compounds. Every under-sink dual-stage and reverse-osmosis filter showed near-complete removal across the board. Standalone carbon filters (pitcher, refrigerator, faucet-mount) were all over the map: long-chain PFAS got pulled down about 60–70%, short-chain about 40%, with no useful correlation to brand or to how old the filter was. Some expired filters made effluent worse than the tap going in (Herkert et al. 2020). EPA's 2024 rule sets enforceable limits of 4 parts per trillion for PFOA and PFOS — and an aspirational limit of zero, because there is no level the agency considers safe (EPA PFAS NPDWR, 2024).
Microplastics. A pour-through test with PVC and PET fragments spiked into treated water showed that membrane-based filters caught 78–100% of particles, while a pitcher with only granular carbon and ion-exchange resin actually released more particles in the effluent than went in — particles that had built up and then released (Cherian et al. 2023). Reverse osmosis pore size makes physical passage impossible through an intact membrane.
What none of them do reliably: handle a microbiological problem (bacteria, viruses, protozoa). For that you want a UV unit or boiling, or an explicit NSF 53 cyst-reduction certification — a generic carbon filter is not enough.
What you're trading for the convenience of not filtering
The case is built on decades, not days. None of this is the kind of harm you feel; that's the entire point — these are the chemicals that don't announce themselves.
The strongest population signal is bladder cancer. The 24% relative-risk increase from disinfection by-products in men is the kind of number that sounds modest until you multiply it by the roughly 1-in-25 lifetime baseline risk and the 40 years you'd plausibly be drinking the same water (Villanueva et al. 2004). The 2025 meta-analysis added colorectal cancer to the same exposure story and found risk visible at concentrations the EPA currently calls safe (Helte et al. 2025). Neither makes a single person's risk dramatic; both make the case for the cheap end of filtering hard to argue against.
PFAS is the slower-burn version of the same problem. The compounds bioaccumulate — they don't leave your body — and have been linked to kidney and testicular cancer, high blood pressure during pregnancy, thyroid disruption, suppressed vaccine response in children, and elevated cholesterol. The EPA's setting the goal at zero isn't bureaucratic theatre; it's the agency saying the evidence doesn't support any safe threshold (EPA PFAS NPDWR, 2024). If your utility shows detectable PFAS in its report, this is the case that earns the under-sink filter on its own.
Lead's old textbook still reads the same way: no safe threshold, lifelong consequences for kids, cardiovascular drift for adults. The Flint headlines made it feel like a Michigan story; the underlying reality is that any home with pre-1986 plumbing or a lead service line carries some baseline lead exposure that the utility's distribution-side numbers don't fully capture. A certified lead filter on the kitchen tap closes that loop for under $200.
The version of this story people you know are living: the friend who develops a chronic taste for bottled water without ever asking what's in their tap; the family in a 1960s home who never thought to check for lead service lines; the relative in a PFAS-impacted utility who reads the annual notice once and files it. None of them feel anything different. The bill comes due, if it comes due, decades after the choice point — which is what makes the inertia so strong and the case for the cheap fix so quiet.
How to pick
The right filter is whichever one matches what's actually in your water — which means the first step is reading the report you've been ignoring.
A note on certification language: "NSF Certified" or "NSF Listed" (or the WQA Gold Seal, or the IAPMO mark) means a real third party tested the claim. "Tested to NSF standards" is marketing — the manufacturer ran their own bench tests and printed the result. Different things.
The reverse-osmosis demineralisation question
Reverse osmosis strips calcium and magnesium along with the chlorine and the lead — that's the same membrane doing the same thing to everything in solution. The question is whether this matters.
For someone eating dairy, leafy greens, nuts, and beans, the mineral contribution from drinking water is small and the deficit from RO is unlikely to register. For someone whose diet runs lean on those — restricted eating patterns, very low-dairy, or athletes losing substantial minerals through sweat — the gap is worth closing. Two cheap ways: a remineralisation cartridge as the final stage of the RO unit ($30–80, installs in minutes), or mineral drops added to the glass. Eating better closes it for free.
None of the filter approaches has a clinical contraindication in the conventional sense. The relevant cautions are about what filters don't fix: microbiological contamination (well water with coliform issues needs UV or boiling, not a carbon filter), and very hard or iron-heavy water (which can foul carbon and scale an RO membrane fast unless pre-treated). And no filter undoes the lead that's leaching out of a downstream fixture or faucet aerator — installed under-sink, the filter cleans the water on its way past, then a few feet of leaded brass can re-contaminate it.
What most filter guides get wrong
- "Bottled water is cleaner than tap." Most US bottled water is regulated by FDA, tap by EPA, and EPA's standards are stricter and more transparent. Roughly a third to nearly half of bottled water in the US is filtered municipal tap water sold back to you. The bottle itself contributes microplastic and, depending on resin, bisphenol exposure.
- "My pitcher removes everything harmful." A standard Brita or Pur pitcher is certified for chlorine taste and odour only. Lead, PFAS, fluoride, nitrate, and most pharmaceuticals pass right through unless the cartridge is specifically certified to NSF/ANSI 53 for those contaminants. The Brita Elite cartridge is — the standard one isn't.
- "Boiling cleans my water." Boiling kills microbes and drives off chlorine. It concentrates lead, nitrate, and PFAS — water evaporates, solutes stay behind. If you're worried about chemical contaminants, boiling makes the problem slightly worse.
- "Reverse osmosis is unhealthy because it strips minerals." Overstated for someone eating a varied diet; the case sharpens for low-dairy diets or where water is the main mineral source (Kozisek 2005). A remineralisation cartridge closes the gap for the cost of a meal out.
- "NSF certification means the filter removes everything under that standard." Each certification is contaminant-specific. A pitcher can be NSF/ANSI 53-certified for cysts and not for lead. Read the data sheet for the named contaminants.
- "Refrigerator filters are basically the same as an under-sink unit." They're not. Most are activated-carbon cartridges sized for taste, not for lead or PFAS. The Duke filter study found refrigerator filters among the most variable performers (Herkert et al. 2020).
Where filtering quietly stops working
The data on certified filters in lab tests and the data in real homes don't tell the same story. A meta-review of NSF/ANSI 53 lead-reduction performance found 99% success in field-deployed filters; the same standards in laboratory tests, deliberately stressed, ran at 61% (Tang et al. 2023). The translation: filters that get the conditions they were designed for work reliably; filters used wrong don't.
The dominant failure mode is the simplest: expired cartridges. Activated carbon has a finite number of binding sites; once they fill, the filter does nothing — and in the case of PFAS, occasionally desorbs what it was previously holding back into your effluent (Herkert et al. 2020). A cartridge that smells fine and pours fast is not telling you it's working.
The second is wrong contact time. A small carbon cartridge running at full faucet flow doesn't give the molecules time to adsorb. This is why under-sink units outperform pitcher cartridges of the same media, and why whole-house chloramine removal needs a deep bed and a slow flow rate.
Third: downstream contamination. A perfectly working filter feeding water through a lead-soldered fixture or a corroded brass aerator can deliver contaminated water out the spout. Replace fixtures if the home is old; periodically unscrew and clean the aerator screen.
Fourth: marketing language that isn't certification. "Tested to NSF standards" means the manufacturer tested it themselves. "NSF Certified" or "NSF Listed" — or the equivalent WQA or IAPMO marks — means a real third party did. Companies pay for certification, which is why uncertified products tend to be the cheap ones. The pattern is consistent: when certification is missing, the bench performance often is too.
What it actually costs
Approximate US pricing as of 2026:
- Pitcher: $20–50 for the jug, $30–100/year in cartridges. Five minutes to set up. Best on chlorine taste; only specific cartridges handle lead.
- Faucet-mount: $30–80 for the device, $50–150/year in cartridges. Screws on in ten minutes. Most are NSF/ANSI 53-certified for lead.
- Under-sink carbon block: $80–300 for the system, $40–100/year in cartridges. DIY install about an hour, or $100–150 for a plumber. The sweet spot for someone who wants reliable lead and disinfection-by-product removal without going to RO.
- Under-sink reverse osmosis: $200–950 for the system, $50–250/year in cartridge and membrane replacements. DIY install two or three hours, or $200–300 for a plumber. The most thorough option.
- Whole-house RO: $4,800–8,000+ installed. Rare for drinking-water purposes alone; usually overkill.
The bottled-water comparison: a family spending $500–2,000 a year on bottled pays back a $500 under-sink RO inside the first year, and every year after that the system runs on $100 of cartridges. The systems last 10–15 years.
RO units send some water down the drain along with the contaminants — older units 3–5 gallons of waste per gallon produced, modern high-efficiency units close to 1:1. At US average water rates, even the wasteful end of the range adds about $5–20/year to the bill. The waste figure gets quoted as a reason not to buy RO; the economics don't actually back that up.
What changes when you start filtering
The honest answer up front: most of the payoff is invisible. This isn't an intervention you feel — it's an intervention whose absence you also wouldn't feel until the diagnosis decades later. Expect the felt experience to be small and the chronic-exposure math to do the real work.
Within days. The water tastes like water, not like a swimming pool. The chlorine smell coming off a fresh glass is gone. People in the household who'd quietly stopped drinking from the tap start using it again — the hydration shift that follows is the only short-term effect you can actually point at.
Within weeks. The kettle stops collecting the slight chlorine off-taste that survived boiling. Coffee tastes cleaner. If you've been buying bottled water, you stop, and the recycling pile shrinks visibly.
Within years. Nothing you'd notice happens. If you switched because of a known lead service line or a PFAS-impacted utility, you're carrying a measurably smaller body burden of those compounds than your unfiltered neighbour — but it's not something the mirror or the bloodwork on a standard physical surfaces.
Across decades. The bladder-cancer risk you weren't going to think about anyway runs a little lower. The colorectal-cancer signal that the 2025 meta-analysis added to the list runs a little lower too (Helte et al. 2025). The lead exposure that would have shaved IQ points off a child or contributed to cardiovascular drift in an adult didn't happen. The PFAS that don't leave your body once they're in — and that your filter prevented from entering — aren't part of your forever chemistry. None of this is dramatic on the level of an individual life; all of it is real on the level of a population.
This is the shape of most chronic-exposure reductions: the payoff lives in the futures that didn't happen, which is exactly why people don't make the change. The reason to make it anyway is that it's cheap, it's set-and-forget, and the downside of being wrong is a few hundred dollars.
Adjacent topics that compound or qualify the filter choice: tap-water testing (home kits versus state-certified lab panels, what to test for in well water versus municipal), bottled water (regulatory gap, microplastic contribution, when it's actually defensible), fluoride in drinking water (contested intervention; only RO and specific carbon blocks remove it), water softening (a different problem — hardness scale, not contamination), shower filtration (chloramine volatiles inhaled during hot showers are part of the disinfection-by-product exposure picture), and the broader PFAS exposure problem beyond drinking water (food packaging, non-stick cookware, textiles, cosmetics) that filtering at the tap only partially addresses.
- — Only reverse osmosis reliably catches microplastics; the filter choice is how you cut that exposure at home.
- — A copper vessel is a folk water-treatment method; a real filter is the dependable alternative.
- — Pick the filter by matching it to whatever your utility report says is actually in the water.
- — Reverse osmosis strips minerals out along with the bad stuff, so RO and remineralizing go together.
- — PFAS from non-stick coatings and industry end up in drinking water; certain filters are built to take them out.
- — The filter logic here applies to your shower head too — and chloramine needs different media than chlorine.
- — Reverse osmosis removes fluoride along with everything else, which matters depending on your view of it.
Substance + claimed effects
The substance is the home water-filtration decision — choosing among the three dominant point-of-use (POU) approaches in residential use: pitcher filters (activated-carbon granules in a gravity-fed jug), in-line activated-carbon block filters (faucet-mount, countertop, or under-sink), and reverse-osmosis (RO) systems (multi-stage with a semipermeable membrane). The claimed consequences cluster around long-term chemical exposure: chlorine and chloramine taste/odour, disinfection by-products (DBPs — chiefly trihalomethanes [THMs] and haloacetic acids [HAAs]), heavy metals (especially lead from corroded service lines and solder), per- and polyfluoroalkyl substances (PFAS), microplastics and nanoplastics, nitrate, fluoride, pharmaceutical residues, and mineral content (calcium, magnesium). The article covers what each approach demonstrably removes, what it does not, the chronic-exposure stakes that motivate the buy, the trade-off RO introduces by demineralising, and the failure modes (uncertified claims, expired cartridges, contact-time too short) that defeat any of the three.
Evidence by addressing question
Mechanism
Three physically distinct removal modes are doing the work:
- Adsorption (activated carbon). The carbon block has an enormous internal surface area (~500–1500 m2/g) of micropores. Organic molecules — chlorine, THMs, many volatile organics, long-chain PFAS — physisorb onto the pore walls. The removal stops when sites saturate, which is why contact time and bed depth dominate performance, and why an expired cartridge can desorb previously trapped contaminants back into the effluent (Herkert 2020). Catalytic carbon is activated carbon surface-modified at high temperature; it chemically decomposes chloramine (NH2Cl) into chloride and nitrogen rather than adsorbing it, which is why chloramine — used by roughly one-fifth of US utilities — needs catalytic media, not standard carbon, for reliable removal.
- Ion exchange. A resin bed (typically Na+ or H+ form) swaps targeted ions out of solution. Pitcher filters often blend granular activated carbon with an ion-exchange resin to capture cationic metals (including some lead). The capacity is finite and exhausts with use.
- Size exclusion (membrane). RO uses a thin-film composite polyamide membrane with effective pore size ~0.0001 µm — about a water molecule's diameter. Water passes under pressure (typically 50–80 psi residential); virtually all dissolved ions, organic molecules, and particles are rejected to a concentrate stream. Microfiltration (~0.1–1 µm) and ultrafiltration (~0.01–0.1 µm) sit between carbon and RO; they capture microplastics and bacteria but not dissolved chemicals.
NSF/ANSI standards codify what a filter must remove, not how. Standard 42 covers aesthetic effects (chlorine taste/odour); Standard 53 covers health effects (lead, cysts, VOCs, asbestos, some PFAS); Standard 58 covers RO systems specifically; Standard 401 covers emerging contaminants (pharmaceuticals, BPA, some pesticides) (NSF/ANSI 53). Manufacturers self-test and pay for certification; the standards are the lingua franca of "this filter actually does what the box says."
Evidence
Chlorine and DBPs. Activated carbon at adequate contact time removes free chlorine essentially completely; standard pitcher and faucet filters are NSF/ANSI 42-certified for chlorine taste/odour reduction. THMs (chloroform, BDCM, DBCM, bromoform) form when chlorine reacts with natural organic matter in source water; the US EPA caps total THMs (TTHM) at 80 µg/L as an annual average (EPA Stage 1 D/DBP Rule 1998). Activated carbon removes THMs by adsorption; a well-maintained carbon block reliably brings TTHM well below the MCL. RO removes them too, both through the pre-carbon stages and the membrane.
The cancer signal that motivates DBP filtration. A pooled analysis of six case-control studies (Villanueva and colleagues, IARC) reported an adjusted OR of 1.24 (95% CI 1.09–1.41) for bladder cancer in men exposed to average TTHM >1 µg/L, rising to 1.44 (1.20–1.73) at >50 µg/L; no association in women, attributed to lower per-capita water consumption and possibly to differences in dermal/inhalation route from showering (Villanueva 2004). The 2025 EHP systematic review and dose-response meta-analysis (29 studies) confirmed positive associations for bladder and colorectal cancer even at exposure levels below current US/EU regulatory limits; risks were strongest in long-duration, high-exposure male cohorts (Helte 2025).
Lead. NSF/ANSI 53 lead certification requires a filter to reduce 150 µg/L challenge water to ≤5 µg/L (updated 2019 from ≤10 µg/L) at both pH 6.5 and 8.5 (NSF/ANSI 53). A meta-review of 1,678 filters across 23 studies found 96% of NSF/ANSI 53-certified filters met benchmark; field-deployed performance was 99%, while laboratory tests (designed to stress) achieved 61% (Tang 2023). By filter form: faucet-mount 98% success, under-sink 88%, pitcher 79%. Standard (non-Elite) Brita pitcher cartridges are not certified for lead reduction; the Brita Elite cartridge is. RO systems certified to NSF/ANSI 58 also reliably remove lead.
PFAS. A Duke / NC State study tested 76 POU systems in central and southeastern NC homes against PFAS (PFOA, PFOS, GenX/PFEAs, others). All under-sink dual-stage and RO units showed near-complete removal of all PFASs evaluated. All other carbon-only filters were variable: long-chain PFAS (≥C6) removed ~60–70%; short-chain PFAS ~40%; no statistical correlation between removal and brand, source-water loading, or filter age. Expired carbon filters sometimes released previously trapped PFAS, increasing effluent concentrations above influent (Herkert 2020). EPA's 2024 final National Primary Drinking Water Regulation set enforceable MCLs at 4 ng/L (parts per trillion) for PFOA and PFOS individually, and 10 ng/L for PFHxS, PFNA, and HFPO-DA (GenX), with the MCL goal for PFOA and PFOS at zero — the agency's position is that there is no safe level (EPA PFAS NPDWR 2024).
Microplastics. The University of Toronto POU evaluation spiked treated drinking water with PET and PVC fragments and nylon fibres (30–1000 µm at 36–64 particles/L) and ran samples at 25–125% of rated capacity. Two devices with microfiltration removed 78–86% of PVC and 94–100% of PET; one device with only granular activated carbon and ion exchange released more particles in effluent than influent at several capacity points, attributed to particle accumulation and release (Cherian 2023). RO membrane pore size (~0.0001 µm) makes microplastic and nanoplastic passage geometrically impossible through an intact membrane; field studies of full-scale RO have detected occasional MP in permeate, attributed to membrane defects. Bottom line: pitcher-style GAC alone is not a microplastic intervention; membrane filtration (UF or RO) is.
Mineral content (the RO-specific concern). RO membranes reject 92–99% of dissolved solids, including beneficial cations (calcium, magnesium, potassium). WHO's 2005 review collated multi-decade epidemiology associating low-mineral water with increased cardiovascular mortality, and documented case reports of acute magnesium-deficiency symptoms (cardiac arrhythmias, fatigue, cramps) appearing within weeks-to-months of switching to RO-only drinking water in populations with low dietary magnesium intake (Kozisek 2005). WHO recommended minimums for drinking water: ~20 mg/L calcium, ~10 mg/L magnesium. Modern under-sink RO systems frequently include a remineralisation cartridge as the final stage; alternatively, the deficit is closed by diet (dairy, leafy greens, nuts) or by mineral drops. In populations with adequate dietary mineral intake the marginal contribution of water is small; the concern sharpens for low-dairy diets, athletes losing minerals through sweat, and pregnancy.
Protocol
The rational decision algorithm is contingent on what's in the source water:
- Read the utility's Consumer Confidence Report (CCR). Required annually; lists TTHM, HAAs, chloramine vs. chlorine, lead-rule-violation status, detected PFAS where measured. If on well water, send a sample to a state-certified lab (~$50–200 for a comprehensive panel).
- Test for lead at the tap if the home pre-dates 1986 (US ban on lead solder) or has lead service lines (most US water utilities now publish a service-line inventory). Even a low utility-side lead number doesn't rule out in-home lead from solder or fixtures.
- Match filter to risk profile. Chlorine taste alone → pitcher or faucet-mount (NSF 42). Lead suspected → NSF 53-certified pitcher (e.g. Brita Elite, ZeroWater), faucet-mount, or under-sink. PFAS-impacted area → under-sink dual-stage carbon block (NSF 53 PFOA/PFOS) at minimum; RO if PFAS levels are high or short-chain PFAS are present. Multiple concerns (lead + PFAS + DBPs + microplastics) → RO, ideally with remineralisation. Chloraminated supply → confirm filter is rated for chloramine (catalytic carbon).
- Replace on schedule. Most carbon cartridges: every 2–6 months or per manufacturer's gallon rating. RO membranes: every 2–5 years. Expired cartridges can underperform or worsen effluent.
Whole-house (point-of-entry) systems sit upstream of all taps. They make sense for chloramine removal (so shower steam doesn't release VOCs) and for hard-water scale management, but are overkill for drinking-water-only concerns. Whole-house RO is uncommon and expensive ($4,800–$8,000+ installed).
Contraindications
Filters themselves carry no clinical contraindications. The relevant cautions are configuration-specific:
- RO without remineralisation in households whose diet is low in dairy, leafy greens, and nuts may incrementally lower calcium/magnesium intake; the WHO acute-deficiency case reports were in low-mineral-diet populations switching abruptly to RO-only water (Kozisek 2005). The downside is closed by a remineralisation cartridge ($30–80) or by dietary attention.
- Filters do not address microbiological risk (bacteria, viruses, protozoa) unless explicitly NSF/ANSI 53-certified for cyst reduction or paired with UV. Well water with coliform contamination needs UV or boiling — a carbon filter alone is not enough.
- Backwash and pre-treatment may be needed for very hard water (RO membrane scales) or high-iron well water (carbon fouls). Pretreatment can add $1,000–3,000 to install cost.
Misconceptions
- "Bottled water is cleaner than filtered." Roughly 25–45% of bottled water in the US is filtered municipal tap water; bottled water is regulated by FDA, tap by EPA, and EPA's standards are generally stricter and more transparent. Bottled water also introduces microplastic and BPA exposure from the bottle itself.
- "My pitcher removes everything." A standard Brita/Pur pitcher is NSF/ANSI 42-certified (taste/odour) only. Lead, PFAS, fluoride, nitrate, and most pharmaceuticals pass through unless the cartridge is specifically NSF 53-certified for that contaminant.
- "Boiling cleans the water." Boiling kills microbes and drives off chlorine, but concentrates lead, nitrate, and PFAS (water evaporates; solutes don't).
- "RO water is unhealthy because it's mineral-free." Overstated in healthy adults eating a varied diet; meaningful in low-mineral-diet populations or where water is the primary mineral source (Kozisek 2005). Remineralisation closes the gap cheaply.
- "All NSF-certified filters remove all contaminants under that standard." Certification is contaminant-specific. A pitcher can be NSF/ANSI 53-certified for cysts but not for lead. Check the data sheet for the specific contaminants listed.
Failure modes
- Expired cartridges. The dominant failure mode. Carbon saturates; thereafter the filter does little, and in PFAS-loaded waters can desorb (Herkert 2020).
- "Tested to NSF standards" ≠ certified. The phrase is unregulated marketing; only "NSF Certified" / "NSF Listed" (or WQA Gold Seal, or IAPMO mark) means third-party verification.
- Wrong contact time. A small cartridge running at high flow can't adsorb effectively; whole-house carbon for chloramine needs a deep bed and slow flow to work.
- Downstream contamination. A certified filter under a lead-soldered fixture or with a contaminated faucet aerator can deliver lead-contaminated water out the spout despite removing it upstream (Tang 2023).
- RO membrane defects. Pinhole defects let microplastics and dissolved contaminants pass; full-scale RO plants detect occasional MP in permeate. Replace per schedule; check TDS meter if quality matters.
Practicalities
Approximate US pricing as of 2026:
- Pitcher: $20–50 for the jug; $30–100/year in cartridges. Setup time: 5 minutes.
- Faucet-mount: $30–80 for the device; $50–150/year in cartridges. Setup: 10 minutes, no tools.
- Under-sink carbon block (single or dual stage): $80–300 for the system; $40–100/year in cartridges; install $0 (DIY ~1 hour) to $150 (plumber).
- Under-sink RO: $200–950 for the system; $50–250/year for filter/membrane replacements (cartridges every 6–12 months, membrane every 2–5 years); install $0 (DIY ~2–3 hours) to $300 (plumber).
- Whole-house RO: $4,800–8,000+ installed; rare for residential drinking-water purposes alone.
RO waste-water ratio: legacy systems waste ~3–5 gallons per gallon produced; modern high-efficiency units approach 1:1. At US average water cost (~$0.004/gallon), even a 4:1 system adds $5–20/year to the water bill — overstated as a deal-breaker.
If currently drinking bottled water, an under-sink RO at $500 install + $100/year amortises within year one against typical bottled-water spend ($500–2,000/year for a family); the system lasts 10–15+ years.
Stakes
The chronic-exposure case is what justifies filtering, and it sits squarely on long timescales — decades of cumulative low-dose exposure to compounds with documented toxicity. The strongest population-level signal is DBPs and bladder cancer: a 24% relative-risk increase in men at typical chlorinated-water exposures (Villanueva 2004), confirmed and extended in the 2025 dose-response synthesis showing risk persists below current regulatory limits and includes colorectal cancer (Helte 2025). PFAS — a class linked to kidney and testicular cancer, hypertension and pre-eclampsia in pregnancy, thyroid disruption, immune suppression, and elevated cholesterol — bioaccumulate; the EPA's MCL-goal-of-zero for PFOA and PFOS reflects the absence of a known safe level (EPA PFAS NPDWR 2024). Lead's neurodevelopmental toxicity in children and cardiovascular effects in adults are textbook and have no safe threshold. Microplastic health effects are still being characterised; the evidence base is younger than the others but the precautionary case is strong.
Payoff
The felt-experience payoff is dominated by taste — drinkable tap water shifts household hydration behaviour, often within days. The chronic-exposure payoff is invisible by design: a small reduction in lifetime risk across several malignancies and chronic conditions, manifesting (or not manifesting) over decades. This is not a high-felt-effect intervention; it is a low-effort hedge with a real but probabilistic upside.
Alternatives
- Bottled water. Worse on most dimensions: more expensive, more variable in quality (less stringent regulation than tap), microplastic exposure from the bottle, plastic waste.
- Distillation. Removes nearly everything (similar profile to RO including demineralisation) but slow, energy-intensive, and a small market.
- UV sterilisation. Microbiological-only; not a substitute for chemical filtration, complementary in some setups.
- Boiling. Microbiological only; concentrates non-volatile contaminants.
- Doing nothing (drink tap as-is). Defensible in many municipal systems if CCR and home plumbing are clean; bad bet in older housing stock, PFAS-impacted regions, or anywhere with lead service lines.
Out-of-scope
Adjacent topics the entry pointers should reach: tap-water testing (the kit market and lab-certified panels), bottled water quality, fluoride (its own contested entry), water softening (different problem — calcium/magnesium scale, not contamination), shower-head filtration for chloramine VOC inhalation, and the broader PFAS exposure problem (food, cosmetics, textiles) that filtering drinking water only partially addresses.
The credibility range
Optimist case. Filtering drinking water is one of the highest-leverage chronic-exposure interventions available — cheap relative to most longevity moves, set-and-forget after install, with a multi-decade payoff. The bladder-cancer signal alone (~24% RR increase in exposed men) justifies action; layer PFAS and lead, and the cumulative argument becomes hard to ignore. The technology is well-characterised, third-party-certified, and the regulatory ratchet keeps tightening (PFAS MCLs of 4 ppt are the canary). For a few hundred dollars and an hour of installation, an under-sink RO closes most of the exposure surface; the only meaningful gotcha (demineralisation) is solved with a $30 cartridge or a varied diet.
Skeptic case. The bladder-cancer association is observational and confounded — heavier water drinkers may differ from light drinkers in ways that bias the estimate; bromide-rich source waters confound; the sex difference is uncomfortable. We have no RCT of filtering vs not on hard health endpoints because the trial is infeasible. Microplastic clinical relevance is unproven; PFAS regulatory thresholds outran the trial evidence; lead in most US homes built post-1986 is a non-issue. Filter markets are riddled with uncertified claims and theatrical marketing; many buyers install systems they don't maintain, which means the field-deployed benefit is smaller than the bench performance. RO demineralisation is real and uniformly under-discussed in sales literature. The honest expected value, for a healthy adult in a modern municipal supply, is modest and not life-changing.
Author's call. The optimist case wins on net but with calibration: this is a sensible hedge, not a game-changer. The case sharpens to "obvious" when any of the following apply: pre-1986 home or known lead service line; PFAS-impacted utility (check the CCR or state PFAS map); high TTHM readings on the CCR; well water; young children in the household. For a healthy adult in a modern home on a good municipal supply, an NSF/ANSI 53-certified pitcher or faucet-mount covers most of the realistic upside at trivial cost; an under-sink RO is the thorough option when PFAS or multiple contaminants are in play. Score evidence at 4 (mechanistic clarity, regulatory backing, replicated observational signal on DBPs; no RCTs on hard outcomes) and controversy at 2 (active but contained debate on demineralisation and microplastic relevance).
Stakeholder + incentive map
- Filter manufacturers (Brita, Pur, Aquasana, APEC, Berkey, iSpring) have commercial incentive to position their product as comprehensive; certification scope often understates this gap. Pitcher brands lean on the chlorine-taste win, which is real but partial.
- Bottled water industry benefits from filtered-tap distrust; safety messaging often frames municipal water as risky in ways the CCR data doesn't support.
- Municipal water utilities have professional incentive to defend tap quality and discourage filter dependence; their CCR data is generally trustworthy but communicates risk minimally.
- NSF International, WQA, IAPMO are independent certifiers; revenue from manufacturer fees creates a structural soft-pressure but the standards themselves are technically rigorous.
- EPA and state environmental agencies set the MCL floor; the 2024 PFAS rule signals that the regulatory tide is moving toward more conservative thresholds, which strengthens the case for self-filtering ahead of utility upgrades.
- Wellness / alkaline-water / "structured water" subcultures push pseudoscientific upgrades (hydrogen water, vortex water); separate from filtration's real evidence base and a noise source in the consumer market.
Population variability
- Geography drives most variance. Source-water quality, treatment-plant practices (chlorine vs chloramine), and pipe age vary widely by utility. The same filter delivers different real-world benefit in Newark, Flint, rural NC, and a new-build in Phoenix.
- Housing age. US homes built pre-1986 (lead solder in plumbing) and pre-1930s (lead service lines) carry baseline lead risk that utility-side data may miss. The Tang 2023 review found field success rates above 99% for certified filters in homes with documented lead exposure (Tang 2023).
- Pregnancy and young children. Lead and DBPs have higher impact at low-dose chronic exposure during neurodevelopment and fetal development; filter thresholds for action should be lower in these households.
- Diet. RO demineralisation matters more in low-dairy, low-leafy-green diets and in athletes losing minerals through sweat (Kozisek 2005).
- Sex differences in the DBP literature. Bladder-cancer association is consistently stronger in men; lower female bladder-cancer baseline, lower water intake per kg, and possibly differences in showering exposure all contribute (Villanueva 2004).
Knowledge gaps
- No RCT of POU filtering vs unfiltered tap on hard health endpoints — and unlikely to be one. The evidence chain remains "filter removes X" + "X exposure causes Y" rather than "filtering changes Y rates."
- Microplastic chronic-exposure dose-response is largely uncharacterised. Removal data is solid; the toxicology is not.
- Short-chain PFAS (PFBS, PFHxA) and ultrashort-chain PFAS are poorly removed by carbon and only partially by RO; the regulatory and removal-technology answer is incomplete.
- Long-term performance variability of certified consumer products in real-world household conditions (variable flow, intermittent use, missed replacements) is under-studied; the Tang 2023 lab-vs-field gap (61% vs 99% lead-reduction success rates) hints at what proper conditions deliver but doesn't fully characterise the median home.
- Real benefit of remineralisation cartridges vs dietary mineral adequacy in mixed diets — the WHO acute-deficiency cases were in specific low-mineral-diet populations; the marginal value for a Western diet is plausibly small but not well-quantified.
Scope. The brief named pitcher filters, activated carbon, and reverse osmosis, plus consequences for chlorine and DBPs, heavy metals, microplastics, taste, and mineral content. All six are covered end-to-end. UV sterilisation and distillation are mentioned as alternatives but not pitched as primary approaches — UV is a microbiological tool that complements rather than substitutes for chemical filtration, and distillation has a tiny consumer market.
Cadence call. Went with as-needed rather than once. The system buy is one-time but the cartridge-replacement trigger is what determines whether the filter works at all; treating the replacement cadence as part of the action seemed more honest than dropping the system into the once-and-done bucket and burying the maintenance in the body.
Longevity score. Sat between 2 and 3. The Villanueva pooled OR of 1.24 and Helte's 2025 confirmation give a real but individually modest signal; PFAS and lead add to the case without independently anchoring a 3. Landed on 2 to avoid inflating a chronic-exposure intervention with no hard-endpoint RCT. The score sharpens to 3 for readers in pre-1986 homes, PFAS-impacted utilities, or with young children — flagged in the body and the credibility-range author's call rather than scored conditionally.
Mood / energy / focus / sleep at 0. No mechanism, no signal in literature. Some wellness marketing claims energy and focus benefits from "purified" or "structured" water — discarded as pseudoscience.
Beauty dimensions at 0. Considered scoring beauty_cumulative at 1 for reduced systemic chemical load → long-term skin effects, but the evidence chain is too thin to justify a non-zero score; the article doesn't dwell on it.
Cost burden at 2. Aggregate score across the filter category — pitcher alone is a 1, whole-house RO is a 4. Weighted toward the typical buyer making an under-sink-class decision.
Audience. Did not scope by age or gender, although the DBP bladder-cancer signal is consistently stronger in men (Villanueva 2004). The intervention applies to everyone in a household, so over-scoping would shrink reach without serving the reader.
Contraindications. None apply from the closed vocabulary. The RO-demineralisation caution is configuration-specific and lives in the article's contraindications section rather than the meta token list.
Separate-entry candidates surfaced during writing.
- Tap-water testing (home kits, lab panels, what to test in well versus municipal supplies) — flagged as a related forward link.
- Fluoride in drinking water — its own contested topic; the filter angle only is mentioned here.
- Shower filtration / chloramine VOC inhalation — a distinct exposure route from drinking; worth its own entry if the catalogue grows the air/breathing intersection.
- The broader PFAS exposure problem (food packaging, cookware, textiles) — water filtering is a partial solution; the full picture is its own piece.
Future links to wire when present. tap-water-testing, bottled-water, fluoride-in-water already added to related; shower-filtration and pfas-exposure-reduction when added.
Hard call on the catalytic-carbon footnote. Considered cutting it from the mechanism section as too in-the-weeds. Kept because roughly a fifth of US utilities use chloramine, and a reader on chloraminated water who buys a standard activated-carbon pitcher gets close to nothing for their disinfectant concern — the gap was worth a paragraph.
Water Filter Types
Buy once, install once. The real work is remembering to change cartridges every few months.
A jug-style filter is under $100/year; a quality under-sink system runs $300–950 up front plus $50–250/year in cartridges.
Removal performance is heavily studied and third-party certified. The bigger uncertainty is whether the individual health payoff is large or modest.
Cuts decades of low-dose exposure to disinfection by-products, lead, and PFAS — small individual risk shifts that add up across a lifetime.
Drinkable tap water shifts hydration habits within days — that's the only short-term win; the chemical-removal benefits are slow and invisible.