PFAS detected vs. over the limit: what the big tap-water numbers count

On November 5 the Environmental Working Group released an analysis of EPA and state testing data that found PFAS in water systems serving an estimated 197 million people, 23 million more than federal testing alone had shown. As the coverage notes, that counts systems where PFAS were detected, not systems over the federal drinking water limits.

Both kinds of number are true, and they answer different questions. A detection tells you a compound is there. The limit tells you whether a system has to do something about it. Here is how far apart they sit in EPA's own national survey, UCMR 5.

From detected to over the limit, in EPA's survey

Of the 3,595 systems that reported a PFAS, about a third averaged over the limit for PFOA or PFOS. By EPA's inventory figures, which count some people twice, the people served drop from roughly 154 million to roughly 55 million.

Why each step narrows

• Most PFAS have no federal limit. UCMR 5 measured 29 PFAS. The 2024 rule set limits covering six of them, and EPA proposed in May to rescind all but the PFOA and PFOS limits. That proposal isn't final, so all six still stand.
• A detection can sit at the limit, not over it. For PFOA and PFOS the reporting level is 4 ppt, the same as the limit. A result of exactly 4.0 is reported but not over. All 28 systems between the third and fourth rows topped out at exactly 4.0.
• Compliance is an average. The rule judges each entry point on a running annual average, so one high sample doesn't make a system out of compliance. 562 systems had a sample over 4 ppt but no entry point averaging over it.
• The survey isn't every system. UCMR 5 was built to sample every system serving 3,300 people or more and only a sample of smaller ones. EWG's higher figure adds state testing data that UCMR 5 doesn't include. The roughly 56,000 systems that weren't in the survey owe their first results by April 26, 2027.

EPA ran its own version of the last step, using every PFAS with a limit, not just PFOA and PFOS. It estimates, in its July 2026 data summary, that 7.8% of systems nationwide would have an average over at least one limit.

State by state

Sorted by systems over on average. “Reported a PFAS” is any of the 29 compounds.

What it means for your work

1. If you run a system, look at your entry-point averages, not your highest result. A detection, or a single high sample, isn't a violation. Compliance with the limits starts in April 2029, and EPA has proposed letting systems ask for two more years.
2. If you design, build or supply treatment, the “over on average” systems are the ones a rule will push to act. The one-sample group is worth watching, because more samples can move an average either way. Our state market briefs split the two.
3. If you're reading the headlines, ask which number you are looking at: detected, over in one sample, or over on average. They can differ by a factor of three or more.

Look up any system's results on the map, or see which systems already have state PFAS funding.

As of November 7, 2026. Counts are our arithmetic on EPA's final UCMR 5 file (released August 27, 2026), excluding lithium, which UCMR 5 also measured. A result reported as below the reporting level counts as zero. An entry-point average needs at least two samples and is rounded to one decimal before comparing with 4.0 ppt. These are survey results, not the compliance monitoring the rule requires. People served comes from EPA's water system inventory and can count some people twice where one system sells water to another; 13 sampled systems have no inventory record. We haven't reviewed the state data in EWG's analysis. This is general information, not a compliance determination, and no figure here is a finding that any system violated a rule.

Sources

• WRAL, Forever chemicals found in water serving 197 million as EPA eyes rollbacks (Nov 6, 2026)
• EPA, UCMR 5 data finder and final results
• EPA, PFAS NPDWR implementation