PFAS: what they are, what the evidence supports, and what EPA has proposed

PFAS is one of the most searched health terms that almost nobody can define precisely, and the imprecision is not the public's fault. The acronym covers thousands of distinct anthropogenic fluorinated compounds [s1]. They are grouped together not by what they do in the body but by what they have in common chemically: the carbon–fluorine bond, the strongest single bond in organic chemistry, which is why the class has acquired the epithet "forever chemicals" [s1].

That grouping is useful for regulators and misleading for readers. A statement like "PFAS cause cancer" is a claim about thousands of compounds, most of which have never been tested in a human population. What follows is an attempt to separate the parts of this subject that rest on evidence from the parts that rest on extrapolation.

Where they come from

Decades of use in aqueous film-forming foam, industrial processes and consumer products have produced persistent, highly mobile groundwater plumes that are now detected on every inhabited continent [s1]. Persistence and mobility are the two properties that matter: the compounds do not break down in the environment on any human timescale, and the shorter-chain ones travel readily through soil and water.

Exposure does not arrive by one route. A study of 40 Faroese homes measured PFAS in indoor air and dust and estimated daily intake for 15-year-old children at 0.001, 0.012 and 0.129 ng/kg bw under low, intermediate and high exposure scenarios — roughly 0.2%, 2% and 20% of the European Food Safety Authority's tolerable weekly intake, against about 60% from the average consumption of pilot whale, a traditional part of the Faroese diet [s7]. That is a single, unusual population, and the numbers should not be transplanted elsewhere. The transferable point is structural: diet dominated, but indoor air and dust were not negligible, and their relative share rises as dietary sources fall away [s7].

What the health evidence actually supports

The strongest human evidence comes from a single unusually exposed population: residents of the mid-Ohio Valley, where a cohort of 32,254 highly exposed persons was assembled and followed [s6]. The C8 Science Panel's work on that cohort found probable links between PFOA exposure and kidney cancer, testicular cancer, raised cholesterol, ulcerative colitis, thyroid disease and pre-eclampsia [s6]. Those conclusions were based on disease incidence followed through 2011 [s6].

A mortality follow-up of the same cohort through 2021, published on 30 August 2026, recorded 5,709 deaths and examined 22 causes of death against cumulative PFOA exposure [s6]. It found a positive significant trend for brain cancer, at a hazard ratio of 1.25 (95% CI 1.05 to 1.50), a suggestive but non-significant result for bladder cancer at 1.17 (0.98 to 1.39), and a significant trend for pancreatic cancer among participants over 72 at 1.23 (1.05 to 1.43) [s6]. The authors state plainly that caution is warranted because of multiple comparisons [s6].

Two things follow from this that are easy to lose. The first is that the evidence base is narrower than the volume of public discussion implies: it rests heavily on one contaminated community, at exposure levels far above the general population's. The second is that "probable link" is a specific and limited claim. It is not a demonstration of cause, and it was reached for a particular compound — PFOA — not for the class.

What the 2024 rule did

The first legally binding US drinking water limits for PFAS were promulgated in 2024, at 4 ng/L for PFOA and PFOS [s1]. That rule, the PFAS National Primary Drinking Water Regulation, also set provisions covering four further substances: perfluorohexane sulfonic acid (PFHxS), perfluorononanoic acid (PFNA), hexafluoropropylene oxide dimer acid and its ammonium salt — the GenX chemicals — and mixtures of those three plus perfluorobutane sulfonic acid (PFBS) [s2].

Four nanograms per litre is an extremely low number. It is roughly the reason the compliance and treatment questions are as difficult as they are: conventional sequestration technologies — granular activated carbon, ion exchange and high-pressure membranes — concentrate PFAS rather than destroying it, generating residual streams that then have to go somewhere [s1].

What EPA has proposed since

Three proposals and one draft guidance are currently in play, and none of them is settled. All are proposals open to public comment, not final rules.

On 20 May 2026 EPA proposed to rescind the regulatory determinations for PFHxS, PFNA, HFPO-DA and the four-substance mixture, along with the associated provisions in the 2024 rule that are exclusive to them — including the final maximum contaminant levels that would have required monitoring and, where necessary, treatment by public water systems [s2]. The stated rationale is procedural rather than toxicological: EPA says the action is necessary to correct what it calls the unlawful procedure under which the regulations were promulgated, because the agency proposed and finalised the regulatory determinations and the regulations simultaneously and in tandem, and under what it describes as the best reading of the statute it is not authorised to do both at once [s2].

The same day, EPA proposed a federal exemption under Safe Drinking Water Act sections 1416(f) and 1450(a)(1) that would extend the compliance dates for the PFOA and PFOS maximum contaminant levels from 26 April 2029 to 26 April 2031 for systems that request it [s3]. The limits themselves are not proposed for change; the deadline for meeting them is.

Moving in the other direction, on 1 July EPA proposed the sixth Unregulated Contaminant Monitoring Rule, which would require public water systems to collect national occurrence data for seven ultrashort organofluorine compounds, including certain PFAS, alongside three pesticide metabolites, 13 semivolatile organic compounds and seven purgeable organic compounds [s4]. Monitoring would cover community and non-transient non-community systems serving 3,300 or more people, plus a representative sample of smaller systems, subject to the availability of appropriations [s4]. Data collected would be publicly available [s4]. Ultrashort-chain compounds are precisely the fraction that existing treatment handles worst, and there is currently no national picture of how much of it is in US tap water.

On 6 July EPA released a draft memorandum on reducing risk from PFOA and PFOS in biosolids — sewage sludge applied to farmland — aimed at wastewater treatment plant operators, landowners and farmers, and state and tribal water agencies [s5]. The agency states explicitly that the draft guidance does not have the force and effect of law and is non-binding [s5]. Comments were open for 60 days, through 4 September 2026 [s5].

How to read the next twelve months

The honest summary is that the science and the regulation are moving on different axes. The scientific direction of travel is toward more compounds, more exposure routes and better measurement — the UCMR 6 proposal is itself an instance of that [s4]. The regulatory direction, in the US, currently includes a proposed narrowing of which substances carry enforceable limits and a proposed delay in when the two remaining limits bite [s2] [s3].

Neither of those proposals has been finalised, and the rescission proposal is argued on administrative-law grounds rather than on a claim that the four substances are safe [s2]. Anyone reading a headline about PFAS rules being "rolled back" or "upheld" over the coming months should check which document is being described, whether it is a proposal or a final rule, and whether the argument is about toxicity or about procedure. On the current record those are three different questions.

Sources

• [s1] From forever to fluoride: A critical review of PFAS occurrence, fate, and the transition from sequestration to destruction-based remediation in groundwater. Science of The Total Environment, 31 August 2026. https://doi.org/10.1016/j.scitotenv.2026.182278
• [s2] Rescission of Regulatory Determinations and Removal of Related Provisions for Four PFAS Substances. US EPA, Federal Register 91 FR 29413, 20 May 2026.
• [s3] Extending the Compliance Deadline for the PFOA and PFOS Maximum Contaminant Levels. US EPA, Federal Register 91 FR 29425, 20 May 2026.
• [s4] Revisions To Establish the Sixth Unregulated Contaminant Monitoring Rule (UCMR 6) for Public Water Systems. US EPA, Federal Register 91 FR 39952, 1 July 2026.
• [s5] Draft Guidance for Reducing Risk From PFOA and PFOS in Biosolids. US EPA, Federal Register 91 FR 41020, 6 July 2026.
• [s6] Mortality study of a PFOA-exposed cohort in the mid-Ohio Valley. Environment International, 30 August 2026. https://doi.org/10.1016/j.envint.2026.110492
• [s7] Contribution of Indoor Air and Dust to Children's Per- and Polyfluoroalkyl Substance Exposure in the Faroe Islands. Environmental Science & Technology, 30 July 2026. https://doi.org/10.1021/acs.est.6c02003