Water safety

Is tap water safe to drink in the US?

Between 2023 and 2025 the federal government tested 10,313 public water systems for PFAS. The results are now complete, and they answer the question better than any general reassurance can — but not in the direction most people expect.

What “safe” actually means in US drinking water law

The word does more work than most people realize. When a utility says its water is safe, it is making a narrow and specific claim: that the substances the Environmental Protection Agency regulates were measured, and the averages came in under the legal limits. It is not claiming the water contains nothing.

Those limits are called Maximum Contaminant Levels, and they are set for around 90 substances. A limit is not a biological threshold below which a compound becomes harmless. It is the product of health science and a feasibility judgement — what the EPA considers achievable given available treatment technology and cost. The agency publishes a separate, non-enforceable goal for each contaminant representing the level at which no known health effects occur. For PFOA and PFOS, that goal is zero.

So “compliant” and “contains nothing of concern” are different statements. A system can be entirely lawful and still put measurable quantities of industrial chemicals through your kitchen tap. This is the gap the 2023–2025 testing program was designed to measure.

What the federal testing program found

The Fifth Unregulated Contaminant Monitoring Rule — UCMR 5 — required public water systems serving more than 3,300 people, plus a representative sample of smaller ones, to test for 29 PFAS compounds and lithium. Sampling ran from January 2023 to December 2025 and EPA completed publication of the results this year. The dataset covers 10,313 systems serving 305 million people: effectively everyone in the country on municipal water.

We processed the complete set of 1,992,002 individual results. Three findings stand out, and the second is the one nobody expects.

1. A majority of systems had something

6,203 of 10,313 systems — 60.1% — detected at least one monitored substance above the laboratory reporting limit. Most of those detections were below any federal limit, and many were of compounds that have no federal limit at all. But the baseline assumption that PFAS contamination is an unusual, localized problem does not survive contact with the data. Detection is the majority condition.

2. Larger systems were worse, not better

This runs against intuition. Big-city utilities have larger budgets, more sophisticated treatment and full-time laboratory staff. They also, it turns out, have considerably more PFAS.

A system serving more than 100,000 people was more than three times as likely to record a sample above a health-based limit as one serving fewer than 3,300. The likeliest explanation is not that large utilities treat water badly. It is geography and history: large systems draw from rivers and aquifers near the airports, military installations, manufacturing plants and landfills where PFAS entered the environment in the first place. Population and industry share the same map.

There is a second, less discussed factor. Large systems have more entry points and therefore submit more samples, so they have more opportunities to record a high one. The effect is real but does not account for a gap this wide.

3. Where you live matters more than almost anything else

Aggregated to state level, the spread is extraordinary. In New Jersey, 58.2% of tested systems recorded a sample above a federal limit. In Arkansas, one system did — 0.6%.

Some of this reflects genuine contamination geography — the industrial corridor of New Jersey, the Cape Fear river basin in North Carolina. Some reflects how aggressively individual states sampled and which systems they prioritized. Both readings support the same practical conclusion: a national average tells you very little about your own kitchen.

One further pattern is worth recording. Groundwater systems detected PFAS slightly more often than surface water systems — 64.0% against 56.3% — but surface water systems exceeded a limit more often, 19.0% against 14.6%. Groundwater picks up low-level contamination more widely; surface water, when it is contaminated, tends to be contaminated harder.

The limits, and what changed in 2026

In April 2024 the EPA finalized the first enforceable federal drinking water limits for PFAS:

Two regulatory changes are in motion, and both are proposals rather than settled law. On 18 May 2026 the EPA proposed rescinding the limits for PFHxS, PFNA, HFPO-DA and the Hazard Index for mixtures; the public comment period closed on 20 July 2026. Two days after that proposal, on 20 May 2026, the agency proposed extending the compliance deadline for the PFOA and PFOS limits from 26 April 2029 to 26 April 2031 for systems that apply and qualify. Systems measuring at or above 12 ppt would have to put interim control measures in place during any extension.

The PFOA and PFOS limits of 4.0 ppt are not part of the rescission proposal. They stand.

Why these compounds are regulated at all

The limits are set in parts per trillion, which is an unusual unit in drinking water regulation and reflects two properties of this class of chemicals.

The first is persistence. The carbon–fluorine bond is among the strongest in organic chemistry, which is precisely why PFAS were useful in non-stick coatings, firefighting foam and waterproof fabric, and precisely why they do not break down in the environment or in the body on any human timescale.

The second is the weight of the health evidence. In 2023 the International Agency for Research on Cancer classified PFOA as carcinogenic to humans (Group 1), citing renal cell carcinoma and testicular cancer, and PFOS as possibly carcinogenic to humans (Group 2B) on mechanistic evidence. Alongside cancer, the EPA's assessments associate exposure with decreased antibody response to vaccination, increased cholesterol, liver enzyme changes, reduced infant birth weight and pregnancy-induced hypertension.

It is worth stating plainly what this does and does not imply. These are population-level associations at a range of exposures, and a hazard classification describes strength of evidence, not magnitude of risk. Drinking water slightly above 4.0 ppt is not comparable to an occupational exposure, and nobody should read a number on this site as a diagnosis. What the evidence supports is the modest, practical conclusion the regulation itself reflects: less is better, and it is worth knowing your number.

What this testing did not cover

UCMR 5 measured 30 specific substances. A result of “nothing detected” is a statement about those 30 and nothing else. Several of the most consequential drinking water risks in the United States sit entirely outside it:

• Lead, which in almost all cases does not come from the treatment plant but from the service line and household plumbing between the main and your tap. This is why lead is regulated at the tap under a separate rule, and why a clean system-level record tells you little about your own house.
• Disinfection by-products, formed when chlorine reacts with organic matter already in the source water — a genuine trade-off against the far larger risk of waterborne pathogens.
• Nitrate, largely agricultural, and acutely dangerous to infants under six months.
• Arsenic, usually geological rather than industrial, and concentrated in specific regions.
• Bacteria and pathogens, monitored continuously under a different rule.

And one exclusion matters more than all of these: if your water comes from a private well, none of this applies to you. Roughly 43 million Americans rely on private wells, which are not covered by the Safe Drinking Water Act, are never federally tested, and are the owner's sole responsibility. If that is your situation, no public dataset can answer the question for you — only a test at a state-certified laboratory can.

What to do with this

1. Find out what your own system measured. One ZIP code, and the answer is specific to your utility rather than to the national average.
2. Read your Consumer Confidence Report. Every community water system must publish one annually, and it covers the regulated contaminants UCMR 5 left out.
3. Treat lead as a separate question. Housing built before 1986 is the risk factor that matters, and it is answered at your tap, not at the plant.
4. Match any filter to what you actually have. Reverse osmosis and certified activated carbon systems reduce PFAS; a standard pitcher filter marketed for taste generally does not. Look for certification to NSF/ANSI 53 or 58 naming the specific contaminant.
5. If you are on a well, test it. Annually for bacteria and nitrate, and once for the broader panel, through a certified laboratory.