A well on one road tests high for nitrate. Within a week, half the street owns a filter pitcher. Almost none of those pitchers do anything about nitrate.
That gap, between what people buy and what their water contains, is the whole subject. Home water treatment is not a quality ladder where spending more gets cleaner water across the board. It is five or six genuinely different machines acting on different molecules, and the one that fixes your problem may be useless for your neighbor’s.
NSF, the standards organization behind most US certification for these products, says so directly. Certification, it writes, “does not mean that a filter, purifier or treatment system will reduce all possible contaminants.”
So the useful order runs backward from the way most people shop. You get a test result. The result names a specific contaminant and a specific number. Then you go looking for a technology proven against that contaminant.
Start from the result, not from the shelf
A water test result is a pair: a substance and a concentration. That pair is the only input that makes a treatment decision possible, which is why article 695 in this library covers testing and this one assumes you have a result in hand.
What matters next is which family your contaminant belongs to, because the families behave nothing alike inside a device.
Dissolved ions. Nitrate, arsenic, lead, chloride, sodium, uranium. Charged particles in solution, far smaller than anything a paper filter catches, and they do not evaporate. Removing them means trading them for other ions, pushing water past them through a membrane, or boiling the water off and leaving them behind.
Organic molecules. Chlorination byproducts, solvents, fuel components, many pesticides, pharmaceutical residues. These stick to carbon surfaces, which is why one cheap technology handles a surprising share of them.
Living things. Coliform bacteria, E. coli, Giardia and Cryptosporidium cysts, viruses. Strained out if the pores are small enough, or inactivated with ultraviolet light.
Nuisance minerals. Hardness, iron, manganese, sulfur smell. These rarely threaten health, but they ruin fixtures, stain laundry and taste bad, and they are what homeowners notice and act on first.
No single media handles all four families. A device excellent at one is typically indifferent to the others, and that is not a flaw in the product. It is what the product is.
What each technology does to the water
Activated carbon, which grabs molecules and then fills up
Carbon works by adsorption: dissolved molecules stick to the vast internal surface of carbon granules or a carbon block. It is in nearly every pitcher, faucet mount and refrigerator cartridge sold.
Carbon reliably reduces chlorine taste and odor plus a long list of organic chemicals. Blocks certified for health claims can also reduce lead, cysts, asbestos, benzene, carbofuran and turbidity, the exact group CDC uses as a sample product label in its guidance on choosing a filter.
What carbon does not do: dissolved salts, nitrate and most minerals pass straight through, and it does not disinfect. CDC notes that most pitcher and refrigerator filters are carbon based and are not designed to remove germs at all. Waste is low, just the spent cartridge. Upkeep is that cartridge, on a schedule.
Ion exchange, which trades one ion for another
A water softener is an ion exchange device. NSF’s softener standard, NSF/ANSI 44, describes a cation exchange resin regenerated with sodium or potassium chloride, swapping those ions into the water in place of the calcium and magnesium that make it hard.
Read that again, because it contains a trade most people miss. Softened water is not stripped water. It carries the ions traded in, and in a sodium-regenerated system that means added sodium.
Softening does nothing about germs. CDC states plainly that softeners do not remove parasites, bacteria or viruses. Anion exchange resins are used against nitrate, arsenic and certain PFAS compounds, but those are separate products with separate certifications, not a setting on your softener.
This matters because softening is what people buy. A 2017 survey of Wisconsin well owners in Science of the Total Environment found the most commonly reported treatment was a water softener, and that iron and hardness were the most commonly reported problems. Those are aesthetic complaints. A 2025 study of southern Alabama wells in the Journal of Water and Health put it bluntly: the treatment choices households had made did not protect against the exceedances those households had.
Waste is salt and a brine discharge at every regeneration.
Reverse osmosis, which pushes water through and leaves the rest
Reverse osmosis forces water under pressure through a semipermeable membrane with openings around 0.0001 micron, according to CDC. Almost nothing fits except water.
That gives it the broadest removal list in the house. CDC credits it with removing parasites, bacteria and viruses along with lead, copper, chromium, chloride and sodium, and with reducing arsenic, fluoride, radium, sulfate, calcium, magnesium, potassium, nitrate and phosphorus.
The catch is that the separated material has to go somewhere. The membrane splits incoming water into treated water and a concentrated reject stream that goes down the drain. How much is wasted depends on the system and the pressure behind it. A 2024 membrane study in Water Environment Research found that higher pressure across the membrane improved both water recovery and the rejection of PFAS compounds, and that pH, salinity and dissolved organic matter all shifted performance. Your product’s data sheet carries its own numbers. An article cannot give you one.
Reverse osmosis is also slow, which is why these systems sit under a sink with a storage tank, and it is indiscriminate: CDC lists fluoride among what it reduces, so a household on fluoridated water is removing something it may have wanted. Upkeep has two tiers. Sediment and carbon prefilters change often and exist partly to protect the membrane. The membrane lasts far longer, and it lasts longer precisely because the cheap filters ahead of it were replaced on time.
Distillation, the oldest method, still working
A distiller boils water, captures the steam and condenses it. Anything that cannot travel as vapor stays behind.
CDC credits distillation with removing parasites, bacteria and viruses plus arsenic, barium, cadmium, chromium, lead, nitrate, sodium, sulfate, calcium, magnesium and many organic chemicals. The failure mode is easy to miss: anything that boils at or below the temperature of water travels with the steam. CDC names some volatile organic compounds, volatile solvents and certain pesticides. NSF’s description of its distillation standard, NSF/ANSI 62, makes the same point about gas-forming chemicals carrying over.
What it wastes is energy and time. CDC says distillation uses more energy and takes longer than other systems. Upkeep is descaling the boiling chamber, where everything you removed is now concentrated, and running the unit regularly, since CDC warns bacteria can grow on the cooling coils of an idle distiller.
Ultraviolet light, which removes nothing at all
An ultraviolet system does not filter. It exposes water to light that damages the genetic material of microorganisms so they cannot reproduce or infect. NSF’s standard, NSF/ANSI 55, splits these units into Class A, intended for water that is not microbiologically safe, and Class B, meant only to reduce nuisance bacteria in water already disinfected. Two very different jobs in similar-looking boxes.
UV does nothing to chemicals, and CDC is explicit about that. It takes nothing out, so inactivated organisms and every particle they rode in on are still in your glass. It also depends completely on clarity: CDC notes units with pre-filtration work better, for a simple physical reason. Anything that shades a microorganism from the light protects it.
Waste is continuous electricity. Upkeep is the lamp, the quartz sleeve around it and the prefilter. A UV system during a power outage is a length of pipe.
A certification is a claim about one contaminant, not a grade
NSF/ANSI standards are written per technology and per claim. Drawing on NSF’s own standards guide, here is what the numbers on a box cover, arranged by how likely you are to need them rather than in numerical order:
- 53 is the health one: reduction of a contaminant with a health effect. It is the number that matters if your test came back with something on it.
- 42 covers aesthetic reduction, chlorine and taste and odor. Most pitchers live here.
- 58 covers reverse osmosis, 62 distillation, and 55 ultraviolet systems in the Class A and Class B split described earlier.
- 44 covers water softeners using cation exchange.
- 401 covers systems verified to reduce one or more of fifteen emerging contaminants, including pharmaceuticals and compounds EPA has not regulated.
- 244 covers supplemental protection against intermittent microbiological contamination on supplies already considered safe, and P231 covers microbiological water purifiers, tested against a 1987 EPA protocol.
- 177 covers shower filters, certified only to reduce free available chlorine.
Look at 177 again. A certified shower filter is certified for exactly one substance. Look at 401 again too: one or more of fifteen. A product can carry that mark having been verified against a single compound on a list of fifteen.
NSF says directly that its standard numbers are not a ranking or a rating system. A product certified under 42 was never tested for anything in 53, and a higher number is not a better product. The number tells you which test bench the product sat on.
So two things matter on a package, and you need both: the standard number and the specific contaminant claim beside it. NSF’s consumer guidance makes that point, and NSF maintains a public database of certified drinking water treatment units, so you can look up a model instead of trusting the carton.
One more detail decides whether a filter stops cysts. CDC distinguishes absolute pore size, where every pore is at or below the rating, from nominal pore size, where the rating is an average and some pores run larger. For parasites, CDC points to an absolute rating of 1 micron or smaller.
Which bucket is this claim in?
A maximum contaminant level is federal regulation, enforceable against a public water system. NSF/ANSI certification is private third-party verification, not a government approval, though it is the verification EPA points consumers toward when it tells people to use a filter certified to remove lead. A government health advisory is guidance without enforcement.
And filters 99 percent of impurities sits in no bucket at all. Ninety-nine percent of which substances, measured how, at what starting concentration, over how many gallons? None of it is specified, because impurities is not a standardized term. A certification to a named standard for a named contaminant answers all of those questions. A percentage with no referent answers none.
One tap or the whole house
Point-of-use treatment handles water at a single fixture, usually the kitchen sink. Point-of-entry, or whole-home, treats water entering the building, typically near the meter on a municipal supply or near the pressure tank on a well. NSF and CDC use the same split, and it maps onto the technologies predictably. Softeners and ultraviolet systems usually go at the point of entry. Reverse osmosis and carbon drinking-water units usually go at a tap.
The trade-off runs both ways. Point-of-use leaves your shower, ice maker and bathroom tap untreated, which matters for contaminants absorbed through skin or inhaled in steam and matters not at all for nitrate. Point-of-entry treats everything, including water you will never drink, and introduces a problem CDC names: a whole-home filter that strips chlorine or other disinfectant can let more germs grow in the plumbing downstream.
EPA researchers tested both configurations when they evaluated point-of-use and point-of-entry units using reverse osmosis and granular activated carbon against PFAS in Colorado, published in AWWA Water Science in 2019. Placement is not academic. It changes which contaminants you have addressed.
Maintenance is the product, not an accessory
NSF describes two different failure modes. A mechanical filter clogs, and the flow rate drops dramatically, which you notice at once. Adsorptive media saturates, and once the surface area is full no more contaminant can be adsorbed. NSF notes that saturation is much harder to detect than clogging.
Consider what that asymmetry means. The failure you can feel is the one that was never dangerous. The failure that matters gives no signal at all. Water keeps flowing at the same rate and tasting the same while the thing you bought it for has stopped happening.
It can get worse than nothing. A 2024 review in Science of the Total Environment on household PFAS treatment reported that improperly maintained activated carbon systems might even raise PFAS levels in treated water, as accumulated compounds come back off exhausted media. That same review found only a handful of peer-reviewed studies measuring PFAS at the tap after household treatment, an honest signal that long-term real-world performance is less documented than the certification marks suggest. EPA makes a narrower version of the point in its lead guidance, warning that a cartridge used past its rated life can become less effective at removing lead.
Two details that cost people money. NSF warns that replacement cartridges are not interchangeable, and that even a small dimensional mismatch can let water travel around the cartridge instead of through it, so the system runs normally while treating nothing. And CDC advises gloves when changing a filter and handwashing afterward, because everything the filter caught is now in your hand.
Adherence belongs in the decision. When researchers gave faucet-mounted filters to 21 households on private wells in West Virginia and Virginia, published in Science of the Total Environment in 2023, the filters significantly lowered total coliform and a long list of metals including lead. Fewer than half the participants intended to keep using them, mostly over flow rate. A device abandoned in month three has a removal rate of zero.
Five questions before you spend anything
- What did the test say, and in what units? Write down the contaminant and the number with its unit. Parts per million and parts per billion differ by a factor of a thousand, and people mix them up constantly.
- Which family does it belong to? Dissolved ion, organic molecule, living organism, or nuisance mineral. That narrows five technologies to one or two before you look at any product.
- Which standard covers that claim, and does this product carry it for this contaminant? The number plus the contaminant, checked in the certifier’s database rather than on the box.
- Where does the water need treating? One drinking tap or the whole building. Then check whether that install is physically and legally possible where you live, since some places restrict softener brine discharge.
- What does a year of upkeep look like, and who will do it? Cartridge count, lamp, salt, descaling, and an honest answer about month nine.
When a filter is the wrong answer entirely
Some water problems are not shopping problems, and treating them as shopping problems delays the step that would have helped.
If your water smells like fuel or solvent, stop. CDC states that water contaminated with fuel or toxic chemicals is not made safe by boiling or disinfection, and a household filter is not the tool either. Call your local or state health department today.
If bacteria appear after flooding or well work, the problem is the well, not the kitchen tap. Articles 695 and 697 in this library cover testing and well systems.
If someone in the household has a weakened immune system, CDC advises talking with a healthcare provider and a water disinfection specialist before selecting a system, choosing one that removes Cryptosporidium, and having someone else change the cartridges, since the person at highest risk should not handle concentrated captured material.
If water changes in taste, smell or color, that is a reason to test rather than to buy. The change is information, and a carbon filter will erase the information without addressing what produced it.
Treat a device sold to make unsafe water safe differently from one sold to improve taste. When a taste filter fails, your water tastes worse. When a Class A ultraviolet unit or a certified purifier fails silently, you drink the failure.
What makes a claim mean anything
The only meaningful claim on a water treatment product is a standard number attached to a named contaminant, and you cannot choose either until you have a test result in front of you.
One practical step: write the replacement date on the cartridge itself with a marker the day you install it, or set a calendar reminder. Every technology here depends on a schedule nobody will remind you about, and the device will keep running, keep flowing and keep looking fine long after it stopped doing the job you bought it for.
Sources
- Centers for Disease Control and Prevention. “About Home Water Treatment Systems.” cdc.gov
- Centers for Disease Control and Prevention. “About Choosing Home Water Filters.” cdc.gov
- NSF. “Standards for Water Treatment Systems.” nsf.org
- NSF. “Home Water Treatment.” nsf.org
- NSF. “Changing Water Filters.” nsf.org
- U.S. Environmental Protection Agency. “Basic Information about Lead in Drinking Water.” epa.gov
- U.S. Environmental Protection Agency. “Basic Information about Your Drinking Water.” epa.gov
- U.S. Environmental Protection Agency. “Protect Your Home's Water.” epa.gov
- PubMed. “MacKeown H, et al. "Removal of perfluoroalkyl and polyfluoroalkyl substances from tap water by means of point-of-use treatment: A review." Science of the Total Environment, 2024. DOI: 10.1016/j.scitotenv.2024.176764.” doi.org
- PubMed. “Patterson C, et al. "Effectiveness of point-of-use/point-of-entry systems to remove per- and polyfluoroalkyl substances from drinking water." AWWA Water Science, 2019. DOI: 10.1002/aws2.1131.” doi.org
- PubMed. “Patton H, et al. "Faucet-mounted point-of-use drinking water filters to improve water quality in households served by private wells." Science of the Total Environment, 2023. DOI: 10.1016/j.scitotenv.2023.167252.” doi.org
- PubMed. “Ma Q, et al. "Evaluation of commercial nanofiltration and reverse osmosis membrane filtration to remove per- and polyfluoroalkyl substances (PFAS): Effects of transmembrane pressures and water matrices." Water Environment Research, 2024. DOI: 10.1002/wer.10983.” doi.org
- PubMed. “Malecki KMC, et al. "Private-well stewardship among a general population based sample of private well-owners." Science of the Total Environment, 2017. DOI: 10.1016/j.scitotenv.2017.05.284.” doi.org
- PubMed. “Millner S, et al. "Drinking private well water: Groundwater quality and management of wells in southern Alabama." Journal of Water and Health, 2025. DOI: 10.2166/wh.2025.380.” doi.org