PFAS turned up more often in big-city water systems than in small towns

Why does a big city find PFAS more than a small town does?

Because of what comes up out of the ground underneath it. Water systems serving more than 10,000 people turned up PFAS at 42.4 percent of them. Systems serving 10,000 or fewer came in at 28.4 percent. Your own supplier is in one of those two piles.

Picture what we sat down with. One plain text file, 1.9 million rows long, one row for every result a laboratory sent in, and not a word anywhere in it telling you what any of that adds up to. Open it and there’s no story in there. It looks like a phone book.

EPA hands that file out to anybody who asks, so rather than lift a figure out of somebody’s summary we split it by system size and counted it ourselves, which is the dull half of this job and the half that decides everything after it.

Of 10,297 water systems with a result, 3,539 had at least one of 29 different PFAS at or above what a laboratory is even allowed to call a finding. All 29 are manufactured chemicals, made on purpose. Somebody put every one of them where it was found. That total matched the figure the agency prints in its own summary, so we knew our arithmetic was clean.

Then a better question started, and it’s the one worth knowing the answer to. Is big-city water dirtier, or only watched harder?

Does a detection mean your water is unsafe?

No. It means a laboratory saw something and was allowed to write it down.

Each test in that file has a floor under it. Below that floor a lab may not put a figure on paper at all, because instruments get shaky down there and a guess is worse than silence. For PFOA and PFOS that floor is 4 parts per trillion. EPA sets it on what machines can do, and its own glossary says these floors “are not associated with health effects information.”

Four parts per trillion is a quantity almost nobody can picture, and that’s most of the point. Laboratories spent years learning to see that far down, and what a machine can see is a different question from what will hurt you.

Two of those 29 chemicals make the point by themselves. PFBA showed up at or above its floor in 5,256 results and PFHxA in 5,899, and neither has ever come back above the health figure it gets screened against. Thousands of findings, no alarm attached to any of them.

Measured against the legal limit instead, that same size pattern shows up again, gentler. Big systems had a sampling point averaging above a PFAS limit 15.3 percent of the time. The middle band came in at 8.6 percent and small systems at 7.3. Nationwide it works out to 8.0 percent.

One caution rides along with those, printed in bold in the source. Results here “do not indicate compliance or noncompliance” with a legal limit, because compliance runs on a separate year-long average. Nothing in those figures says any utility broke a law. If your own supplier turns up in that column, nobody has accused it of anything.

Do big utilities simply test more taps?

That was our first suspicion, and it deserved a test rather than a shrug.

A big utility hands over more sampling points than a small one. 3.42 each on average, against 1.97, which is three or four taps in a city and two in a small town. More taps means more chances for something to show up, and any whole-system rate rewards that.

So we stopped counting systems and counted taps. Big systems gave 15,004 sampling points, 26.7 percent of them with a PFAS finding. Small systems gave 11,658 points at 20.6 percent.

Then we cut it a second way, keeping only utilities that reported one sampling point and no more, so every system got a single shot at finding anything. Big ones came in at 32.8 percent. Small ones at 25.3.

That gap narrows both times, from 14 points down to 6 and then 7. It never closes, and it doesn’t change direction either.

Why does the source of the water decide it?

Split those sampling points by where a system gets its water, and half of this flips over.

Read that top row twice. Out of rivers and lakes, small systems came out slightly worse than big ones, which is the opposite of everything above.

Every bit of that size gap is in the bottom row, in water pumped up out of the ground.

Think about which of those two your own tap runs on. Most people have never been told, and it turns out to be the line this whole thing splits along.

Source mix can’t explain it away either. Wells account for 74.0 percent of big-system sampling points and 74.2 percent of small-system ones, so both groups draw from nearly the same mix of rivers and wells.

Geography was our next suspect, since big utilities and small ones aren’t sprinkled evenly across this country. We ran it again one state at a time, across 38 states and territories with at least 50 sampling points in each group. Big systems came out higher in 28 of them and lower in 10. Minnesota spread wide, 53.2 percent against 25.0. New Jersey ran backwards, 51.0 against 54.7. We picked those two as examples of each direction rather than as records.

Does anything run the other way?

One thing in that file isn’t a PFAS at all, and it’s why we believe the rest of this.

Lithium got measured in the same round, at the same systems, on the same sampling trips. Nobody manufactures lithium into drinking water. It leaches out of rock, which makes it the one contaminant in this whole file that no factory ever put there.

If big utilities simply produce more findings, because their crews sample harder or their labs work faster, lithium would lean their way too. It leans the other way. 28 percent of small systems came back above the health figure lithium gets compared against, and 24 percent of big ones.

Same file. Same crews. Opposite direction.

Say the caveat out loud, because it matters. That’s a share above a health figure rather than a detection rate, so it isn’t like-for-like with anything above it. Treat it as a check that went hunting for bias and failed to find any.

What about the smallest towns, and everybody on a well?

Most of them were never sampled at all, and that’s written into the rule rather than hidden by it.

Any system serving 3,300 people or more had to take part. Below that line, only a national sample had to, so plenty of small towns simply sat this out.

Plans for the next round put a size on that hole. It expects 800 randomly chosen systems serving fewer than 3,300 people, all 5,155 in a middle band, and all 4,599 above 10,000. That comes to 10,554 systems, out of more than 148,000 public water systems in this country.

Underneath all of it, a whole population goes unmonitored. More than 43 million Americans, about 15 percent of us, drink from a private well. No round of this has ever reached a single one of them. If that’s your water, this file says nothing about you, no federal limit applies to you, and no annual sheet turns up in your mailbox to say whether anybody has checked.

Nobody bills you for that, nobody reminds you about it, and the only person who will ever find out what comes out of a private well is whoever decides to pay a laboratory to go and look.

So the quietest stretch of this map is also the least measured stretch of it.

Is the PFAS limit changing?

Partly. Two of six figures in that 2024 rule look safe. Four are up for deletion.

April 2024 put a legal limit of 4.0 parts per trillion on PFOA and PFOS, 10 parts per trillion each on three more, and a combined limit on mixtures. Utilities have until 26 April 2029 to meet it.

In May 2025 EPA said it would keep the PFOA and PFOS limits and reconsider everything else. A year later came two proposals: one letting systems that ask push their deadline out to 26 April 2031, another rescinding those four remaining figures.

Both are proposals, and comments on the deadline closed in July. Until somebody finishes one, that 2024 rule is still the law, unchanged, and April 2029 is still the date your water company is working to.

Say that plainly, because a headline about a rule being rolled back sounds as though something has already changed. Nothing has. A proposal asks whether to change a rule, and until it becomes final the limit your utility must meet is the one written in April 2024.

So the useful question is what your own supplier has already found and reported, rather than which way the rule goes. Those results are in a public dataset today and will stay there whichever way it ends up.

What does this tell you about your own tap?

It says PFAS turns up more in water pumped from under big systems. It doesn’t say small-town water is cleaner, and neither figure tells you what came out of your own tap this morning.

Every check we could throw at that pattern left it standing. Count taps instead of whole systems, and the gap appears again. Restrict it to utilities with a single sampling point, and the gap is there. Go one state at a time, and it turns up in 28 of 38. Split by source, and it collapses into wells. Swap in a contaminant that comes out of rock instead of a factory, and it reverses.

What none of it explains is why. That file carries no field for where any of this came from, so a well running higher under a big utility is a pattern we can show you and can’t account for.

It does say plenty about who got looked at, though. Utilities most likely to find PFAS are big ones. Small towns and every private well were largely left out. That’s a fact about who got tested, rather than about what’s in your pipe.

If your water comes from a community system that took part, its results had to appear in your annual water quality report, average and range both. EPA’s UCMR 5 Data Finder shows you the same rows we counted, filtered down to your own utility. Anybody can go and look, which is the one good thing about a file this size being public.