Step one, and it's free

If you are on a community water system, your utility is required to send you a report every year. It is called a Consumer Confidence Report, and it must reach customers by July 1 each year.

By regulation it has to tell you:

• The lake, river, aquifer or other source of your drinking water
• A brief summary of the risk of contamination of that source
• The regulated contaminants found in your local drinking water
• The potential health effects of any contaminant detected in violation of a health standard
• What the system did to restore safe water
• An educational statement for vulnerable populations about avoiding Cryptosporidium
• Educational material on nitrate, arsenic or lead where applicable
• Phone numbers for more information, including EPA's Safe Drinking Water Hotline, 1-800-426-4791

Most people throw this away. It is the single most useful document in this entire subject, because it is specific to your tap and it costs nothing. EPA hosts an online tool to find your local report, and if that fails, you can call your water supplier directly.

If you are on a private well, none of this applies to you. Private wells are not regulated by the federal government under the Safe Drinking Water Act, and are not regulated by most state governments either. Well owners are responsible for their own water. CDC's advice is to test at least once a year for total coliform bacteria, nitrates, total dissolved solids, and pH, plus anything else of local concern, using a state-certified laboratory. Your local health department can point you to one. (That annual testing advice is CDC's; the EPA private wells page we retrieved does not give a testing frequency.)

How to read the numbers you find

Three terms, and the difference between them matters.

• MCL: Maximum Contaminant Level. EPA defines it as the highest level of a contaminant allowed in drinking water. These are enforceable.
• MCLG: Maximum Contaminant Level Goal. EPA defines it as the level below which there is no known or expected risk to health. These are goals, set with a safety margin, and are not enforceable.
• TT: Treatment Technique. A required process intended to reduce the level of a contaminant. Used where measuring the contaminant directly is impractical.

Reading an MCLG as a danger threshold is the most common misreading of this framework. Several MCLGs are zero. That does not mean any detectable amount is harmful; it means the goal was set at zero.

There is also a second, separate set of secondary standards, 15 of them, covering things that affect taste, colour and appearance. EPA does not enforce these. They are guidelines for managing aesthetic problems. If your water smells of chlorine or leaves orange marks, that is usually a secondary-standard issue, which is to say a nuisance rather than a hazard.

Some values worth having:

On fluoride, three different numbers circulate, and they are three different things: 4.0 mg/L is the enforceable limit, 2.0 mg/L is a non-enforceable secondary standard, and 0.7 mg/L is the Public Health Service's recommended level for community water fluoridation (a recommendation, not a limit).

Lead, the complicated one

Lead does not have an MCL. Its MCLG is zero, and it is handled through a treatment technique with an action level, a number that, when exceeded, triggers required steps by the utility rather than being a violation on its own.

The action level is currently mid-transition, so any number you see needs a date attached. The Lead and Copper Rule Improvements, finalised October 8, 2024, revised the action level from 0.015 mg/L to 0.010 mg/L, with a compliance date of November 1, 2027. Until then, systems may follow the earlier versions of the rule. So 0.015 mg/L is what systems are held to today, and 0.010 mg/L is codified and takes effect November 1, 2027. Both figures are correct with their date and misleading without it.

The rule also requires utilities to replace lead service lines within 10 years, at an average annual rate of at least 10%.

Every water system was required to develop an initial service line inventory and submit it to the state by October 16, 2024. The inventory must be publicly accessible, and systems serving more than 50,000 people must post it online. That makes it the most actionable thing here: if you want to know whether the pipe into your house is lead, that inventory is where to look, and your utility is the place to ask. EPA also publishes a step-by-step guide called Protect Your Tap for identifying lead pipes in a home.

EPA's household advice on lead: use only cold water for drinking, cooking and making baby formula; use a filter certified to remove lead; run the water before drinking to flush the pipes; and regularly clean your faucet's screen.

EPA says only that the filter should be certified to remove lead. It does not attach a standard number to that advice, so neither will we. Boiling water does not remove lead. EPA says so directly, and adds that boiling and chemical disinfection will not destroy heavy metals, salts, or most other chemicals, so boiling is a tool for germs, not for anything dissolved.

Replacing a lead service line can temporarily raise lead at the tap, which is why the rule requires written notice to customers. Improvement is not always immediate.

PFAS, where the status is unresolved

In April 2024, EPA finalised national drinking water limits for certain PFAS, the so-called "forever chemicals." PFOA and PFOS were each set at 4.0 parts per trillion, with goals of zero. Four other entries (PFHxS, PFNA, GenX, and a Hazard Index for mixtures) were also included.

Then, on May 18, 2026, EPA proposed two changes: one to let systems request until 2031 to comply, and one to rescind the regulations for PFHxS, PFNA, GenX and the Hazard Index.

As of September 19, 2026, we could not determine whether either proposal has been finalised. EPA's own page had not been updated since it announced the proposals.

The defensible statement is that the PFOA and PFOS limits of 4.0 parts per trillion were set in April 2024, that EPA proposed in May 2026 to rescind the limits for four other PFAS and to allow systems more time, and that the outcome should be re-checked at EPA's website. If you are reading this some time after publication, please do check. This is the section most likely to be out of date.

Separately, EPA has stated that its 2022 interim health advisories for PFOA and PFOS no longer reflect the best available scientific information. If you encounter those older, far smaller numbers, EPA has disowned them.

Matching technology to contaminant

Now for the comparison. We are going to be unusual here and leave gaps where we could not verify something, rather than filling in the table from general knowledge.

Activated carbon. The pitcher, the faucet attachment, the fridge filter. CDC's description is blunt: their main purpose is to improve the taste and smell of water rather than its safety. The associated standard is NSF/ANSI 42, which covers taste and odour. Most home carbon filters of this type do not remove germs. Some carbon products carry additional certifications for specific contaminants. If one does, it will name the standard and the contaminant on the package; if the package does not name them, assume it is a taste-and-odour filter. We could not retrieve a source describing how common those additional certifications are, so we are not estimating.

Reverse osmosis. It removes germs including viruses, and CDC names lead, copper, chromium, chloride and sodium among what it removes. CDC says it "may reduce" arsenic, fluoride, radium, sulfate, calcium, magnesium, potassium, nitrate and phosphorus. That hedge is CDC's and we are keeping it. It is associated with standard 58. CDC's list does not mention PFAS, and we found no agency source for RO and PFAS, so we are not making that claim.

Membrane filters, by pore size. This is where the label details matter. For parasites, CDC specifies an absolute pore size of 1 micron or smaller, or microfiltration/ultrafiltration/nanofiltration/RO, or certification to standard 53 or 58. For bacteria, it specifies an absolute pore size of 0.3 micron or smaller. For viruses: nanofiltration or reverse osmosis only.

One more detail matters on a label: "nominal" pore size means an average. Some pores are smaller and some are larger, so larger contaminants can pass. Only an absolute rating is protective.

Distillation. CDC names standard 62 for distillation systems but gives no removal detail on the page we could retrieve. We are not going to describe what distillation does or does not remove on that basis.

UV, ceramic, ion exchange and water softeners. We could not retrieve an agency source describing any of these. The common claims are plausible. They are also unsourced here, so we are leaving them out rather than repeating them. A softener in particular is a hardness appliance; do not assume it is doing anything about contaminants.

Carbon block versus granular carbon. No source we retrieved draws this distinction, so we cannot tell you there is a performance difference.

Reading a certification without being fooled

"NSF certified" on its own tells you nothing, and knowing that is the most useful consumer skill in the whole subject. Third-party certification is granted against a specific standard for a specific claim. CDC names a different standard for each job: 42 for taste and odour, 53 for cyst reduction, 58 for reverse osmosis systems, 62 for distillation systems.

A filter certified to standard 42 is certified to improve taste. That is a real certification, and it says nothing whatsoever about lead.

So the question to ask of any box is: certified to which standard, for reduction of which contaminant? That information is either printed on the package or available from the manufacturer's performance data sheet. If it is not, the claim is not a certification claim.

We looked into the scopes of several other standards you will see referenced, including which standard now covers PFAS reduction, and could not retrieve them. We are not going to describe standards we could not read.

Two specific hazards worth stating plainly

Nitrate and infants. This is established, serious, and not a matter of debate. High levels of nitrate or nitrite in drinking water can cause methemoglobinemia, sometimes called "blue baby syndrome," by impairing the blood's ability to carry oxygen. EPA states that infants below six months who drink water with high levels of nitrate can become seriously ill and die. Symptoms include shortness of breath and blueness of the skin.

To keep this in proportion: the MCL of 10 mg/L exists precisely to prevent it, community water systems are tested against it, and your Consumer Confidence Report covers it. The realistic risk sits with infant formula made using untested private well water, particularly near agricultural land or septic systems. Boiling does not help, and EPA is explicit that boiling will not destroy heavy metals, salts and most other chemicals. If you have a well and an infant, test for nitrate before making formula with it.

Cryptosporidium and people with weakened immune systems. CDC's guidance where water safety is uncertain gives three routes: commercially bottled water; water previously boiled for at least one minute and cooled; or a filter designed to remove Crypto, where CDC says the label might read NSF 53 or NSF 58, or state an absolute pore size of 1 micron or smaller.

An ordinary carbon pitcher does not qualify, which is the clearest real-world illustration of why "certified" needs a number.

On boiling generally. EPA says to bring water to a rolling boil for at least one minute, and to boil for three minutes above 5,000 feet. CDC gives the altitude threshold as 6,500 feet. Both were checked this month. This is a real difference between two agencies rather than a typo, so whichever you follow, know which one you are following. Boiling handles pathogenic bacteria, viruses and protozoa, and nothing chemical.

Bottled water is not the escape hatch

Bottled water is regulated by FDA as a food, not by EPA. FDA sets standards of identity, standards of quality, and manufacturing requirements.

You will often read that FDA's bottled water limits are at least as strict as EPA's tap water limits, but FDA's own description does not say that. It says FDA either adopts EPA's standard or finds that the standard is not necessary for bottled water. That is a choice in two directions, not a floor.

Some FDA limits are in fact tighter: the bottled water limit for lead is 0.005 mg/L, well below EPA's tap water action level. Others differ in structure.

Neither agency provides a comparison of safety. No agency source we retrieved states that bottled water is safer than tap water, or the reverse. Anyone claiming either direction is going beyond the record.

One more thing the regulations settle cleanly. The standard-of-identity names in the bottled water rule are: bottled water, drinking water, artesian water, ground water, mineral water, purified water, demineralized water, deionized water, distilled water, reverse osmosis water, sparkling bottled water, spring water, sterile water, and well water.

"Alkaline water," "structured water," "hydrogen water," "ionized water" and "oxygenated water" are not among them. They are not defined water types under this rule, and they do not appear in EPA's drinking water standards either. That is a statement about regulatory standing, which is all we checked. It is not a verdict on any health claim; we did not evaluate that literature.

Keeping a filter honest

Whatever you buy, it has a service life. CDC ties performance directly to maintenance: any filter has to be maintained to keep it working properly and to prevent germs from growing in it, which means changing filters according to the manufacturer's recommendations, wearing gloves while changing them and washing your hands afterwards.

There is laboratory work showing that biofilms do form inside home filter devices, with low flow rates, long stagnation periods and higher temperatures all favouring bacterial growth. Two limits travel with that research: neither study measured whether anyone got sick, and both were small single-device laboratory studies. They support CDC's instruction to change the filter on schedule, and nothing stronger than that.

Where this article stops, on purpose

Everything defensible in this subject is about a device removing a contaminant from water. That is a mechanical claim about a filter: measurable, testable, certifiable.

It is a different thing from a claim about a person. Nothing here suggests that anything you swallow addresses contaminant exposure, and nothing in the sources we used supports that. If you are concerned about a specific exposure (particularly lead in a household with children or a pregnancy), that is a conversation with a doctor, who can decide whether testing is warranted. A filter article cannot assess anybody's risk and this one is not trying to.