Are there "forever chemicals" in your whisky? PFAS, blending water and what changed in 2026

Key takeaways

• The route into a bottle is the water, not the still. A 2025 study in Environmental Science & Technology found PFAS in 95% of 23 US beers tested and reported “a strong correlation between PFAS concentrations in municipal drinking water and levels in locally brewed beer” (ScienceDaily; Redmon et al.).
• Ordinary process filtration is not designed for this. The same study noted that although breweries have water treatment, “they are not designed to remove PFAS” (ScienceDaily).
• Two regimes came into force in 2026. EU monitoring of PFAS in drinking water applies from 12 January 2026 (European Commission), against parametric values of 0.1 µg/L for “Sum of PFAS” (20 substances) and 0.5 µg/L for “PFAS Total” (Measurlabs). In the US, EPA’s May 2026 proposal keeps PFOA and PFOS at 4.0 ppt each while extending compliance to 2031 and rescinding the other four limits (US EPA).
• Purification works; boiling does not. EPA names granular activated carbon, anion exchange, reverse osmosis and nanofiltration as Best Available Technologies, with high-pressure membranes “typically more than 90 percent effective at removing a wide range of PFAS” (US EPA). Minnesota’s health department is blunter about the alternative: “Boiling water will not remove PFAS” (MDH).
• India has no PFAS limit to comply with yet. IS 10500 : 2012 sets no parametric value for PFAS (IS 10500 : 2012), even though PFAS are documented in the Ganges basin at up to 10.2 ng/L for PFBS in river water (Sharma et al.) — which makes the water specification a commercial question before it is a regulatory one.

Almost certainly there is no meaningful quantity, and the reason is worth stating precisely: PFAS enter beverages through water, and the water that reduces a spirit to bottling strength is the most heavily purified ingredient in the bottle. The risk is not in the distillation — PFAS do not volatilise at water’s boiling point, which is why distillation is used to remove them — but in what happens after it, when a cask-strength spirit is diluted with something that may be municipal water in disguise. That is the step worth asking questions about.

What are PFAS, and why is a spirits producer suddenly being asked about them?

Per- and polyfluoroalkyl substances are a family of synthetic compounds built around carbon–fluorine bonds, used since the mid-twentieth century in non-stick coatings, firefighting foams, food packaging and textile finishes. The carbon–fluorine bond is the strongest in organic chemistry, so the molecules do not break down in the environment on any human timescale — hence “forever chemicals.”

They became a drinking-water question because they are mobile in groundwater, detectable at parts per trillion, and now regulated at those levels. Two things then follow for anyone selling a liquid in a bottle. First, the analytical methods exist: EPA Method 533 and 537.1 were written for drinking water, and the 2025 beer study was, on its authors’ account, the first to adapt Method 533 to a finished beverage. Second, once a method exists, a retailer or an importer can ask for the number. That is generally how a contaminant moves from a research topic to a line on a certificate of analysis.

How would PFAS get into a bottle of whisky?

Through the water, almost certainly, and the beer evidence is the clearest available signal.

The study — “Hold My Beer: The Linkage between Municipal Water and Brewing Location on PFAS in Popular Beverages,” by Jennifer Hoponick Redmon and colleagues at RTI International (DOI: 10.1021/acs.est.4c11265) — tested 23 beers and found PFAS in 95% of them, with PFOS and PFOA among the compounds identified. The geography did the explanatory work: beers brewed near North Carolina’s Cape Fear River Basin, an area with documented PFAS pollution, “showed the highest levels and most diverse mix of forever chemicals” (ScienceDaily). The lead author’s framing is the one to carry across to spirits: “The presence of ‘forever chemicals’ in municipal water gets into the beer production process — and the beer itself” (RTI).

Beer and whisky differ in one respect that matters enormously here. Beer is water that was never distilled; a malt whisky passes through a still twice. PFAS are not volatile at the temperatures a still runs at — that is precisely why distillation is an effective removal technology, and why boiling a contaminated water concentrates PFAS rather than driving them off (MDH). A distilled spirit should therefore leave the still essentially clean of them regardless of how the mash water was sourced.

Which leaves reduction. Every route from still to bottle involves adding water back: to fill strength before the cask, and again to bottling strength afterwards. No published survey of PFAS in whisky was available at the time of writing, so this is reasoning from the chemistry and from the beer data rather than from a direct measurement in spirits — but the logic points at one step, and it is the step a producer controls entirely through its water specification.

How much of the bottle is actually water?

More than most drinkers assume, which is what makes the specification consequential.

Take a cask filled and matured to 63.5% ABV and bottled at 40%. Ignoring the small volume contraction that occurs when ethanol and water mix — covered separately in our note on proofing arithmetic — every litre of cask spirit becomes about 1.59 litres of bottled whisky. The added water is roughly 37% of the finished bottle.

The arithmetic for a contaminant is then direct. Whatever concentration the blending water carries appears in the bottle at about 0.37 times that figure, since the spirit side contributes essentially nothing. A blending water sitting exactly at the EU’s 0.1 µg/L “Sum of PFAS” parametric value would put roughly 37 ng/L into the bottle. A purified blending water reporting non-detect at a part-per-trillion reporting limit puts in nothing measurable. The spread between those two outcomes is a procurement decision, not a process one.

What do the rules actually require in 2026?

Four regimes, moving at different speeds and in different directions.

Two features of that table deserve comment.

The first is the gap between the American and European numbers — 4 ppt for a single compound against 100 ppt for a sum of twenty. They are not measuring the same thing, and WHO was candid about why its own figure sits where it does: the provisional values were set recognising that lower levels are unlikely to be achievable globally, particularly in resource-limited contexts. A drinking-water limit is an infrastructure judgement as much as a toxicological one.

The second is India’s blank cell. IS 10500 : 2012 predates PFAS regulation everywhere and sets no limit for it, while the research record shows the substances are present: a survey of the Ganges basin frequently detected 15 PFAS in river water, with PFHxA at 0.4–4.7 ng/L and PFBS up to 10.2 ng/L, and reported that “daily PFAS exposure intakes through drinking water were below safety thresholds for oral non-cancer risk in all age groups” (Sharma et al.). For an Indian producer bottling for export, the operative standard is therefore not the domestic one — it is whichever market’s rules the bottle will land in, a pattern that recurs throughout water quality standards for spirits.

Does purifying water actually remove PFAS?

Yes — and the technologies involved are the ones a blending-water plant already runs.

EPA’s designation of Best Available Technologies for meeting the PFAS limits names granular activated carbon, anion exchange, reverse osmosis and nanofiltration. On the membranes, EPA states that “research shows that these types of membranes are typically more than 90 percent effective at removing a wide range of PFAS, including shorter chain PFAS” (US EPA). On carbon and resin, the agency’s language is about duty cycle rather than efficiency: “GAC can be 100 percent effective for a period of time, depending on the type of carbon used, the depth of the bed of carbon, flow rate of the water,” and “like GAC, AER removes 100 percent of the PFAS for a time that is dictated by the choice of resin, bed depth, flow rate.”

That last point is the operational one. Carbon and resin do not fail gradually in a way you would taste; they saturate and break through. A PFAS claim therefore rests on a change-out and verification regime, not on a flow diagram — the same argument that applies to microbiological control downstream of the membrane.

There is also a pleasing symmetry with what reverse osmosis is bad at. RO rejects by charge and size, which is why it struggles with boric acid and dissolved gases — small, uncharged species, as set out in five things an RO membrane leaves behind. PFAS are the opposite case: large molecules that are anionic at drinking-water pH. They sit squarely in the region where membranes and anion-exchange resins work best, which is why every one of EPA’s four named technologies is a charge- or size-based separation.

What does not work is worth naming too, because it is the folk remedy. “Boiling water will not remove PFAS,” the Minnesota Department of Health states; boiling concentrates them as water evaporates. Nor does a standard domestic filter jug necessarily help — Minnesota advises looking for products certified to NSF/ANSI 53 or NSF/ANSI 58, with the caveat that current certifications “do not yet indicate that a filter will remove PFAS down to the levels EPA has now set for a drinking water standard” (MDH). A carbon jug that removes chlorine, as discussed in our comparison of tap, filtered, mineral and demineralised water, is not thereby a PFAS filter.

What should a buyer ask for?

Five questions separate a water with a PFAS position from a water with a marketing claim.

1. Which analyte list, and to what reporting limit? “PFAS-free” is not a specification. EU Sum of PFAS-20, or EPA’s PFOA and PFOS individually, with a stated limit of quantification in ng/L, is.
2. Which method? EPA Method 533 or 537.1 for drinking-water matrices, or EN 17892 for the EU sum parameter. A number without a method cannot be compared with another number.
3. Tested at which point? Feedwater, post-treatment, or the water as delivered. Only the last one describes what will meet the spirit, and storage and handling sit between the other two and the tanker.
4. How often, and what happens at break-through? For carbon and resin stages, the media change-out interval and the verification test that triggers it.
5. Is the treatment train documented? Reverse osmosis followed by ion exchange or EDI is the standard architecture for demineralised water for spirit dilution, and it happens to be a PFAS train by construction — but only a documented one can be audited.

None of this makes PFAS the most urgent thing in a water specification. Calcium still causes more haze, and microbiology still causes more recalls. What it does is add a parameter that a purifying treatment train handles as a by-product of doing its ordinary job — and that a softened or merely filtered water does not handle at all. That distinction, between water that has had its minerals swapped and water that has had its contents removed, is the one running underneath the whole subject, and we set it out at length in softened water is not purified water.

Glossary

• PFAS — per- and polyfluoroalkyl substances; a family of thousands of synthetic fluorinated compounds, persistent in the environment because of the strength of the carbon–fluorine bond.
• PFOA / PFOS — perfluorooctanoic acid and perfluorooctane sulfonate; the two longest-studied PFAS and the two retained in the US drinking-water rule at 4.0 ppt each.
• Long-chain vs short-chain — a distinction by carbon number that governs treatment: carbon adsorption performs better on long-chain PFAS, while anion exchange and tight membranes handle short-chain compounds more reliably.
• ppt / ng/L — parts per trillion; one nanogram per litre. The unit PFAS regulation is written in, roughly a thousandth of the microgram-per-litre scale used for most trace metals.
• Sum of PFAS — the EU parameter summing 20 named perfluoroalkyl acids from Annex III of the Drinking Water Directive, with a parametric value of 0.1 µg/L.
• Hazard Index — the US mechanism for regulating a mixture of PFHxS, PFNA, HFPO-DA and PFBS together; proposed for rescission in May 2026.
• Best Available Technology (BAT) — EPA’s designation of treatment processes demonstrated capable of meeting a given limit at reasonable cost; for PFAS these are GAC, anion exchange, RO and nanofiltration.
• Break-through — the point at which an adsorptive medium is saturated and the contaminant begins passing into the product water; the governing failure mode for carbon and resin stages.
• EPA Method 533 / 537.1 — validated analytical methods for PFAS in drinking water; Method 533 was adapted by the 2025 beer study for a finished beverage.

Frequently asked questions

Does whisky contain PFAS? No published survey of PFAS in whisky was available at the time of writing. The closest evidence is the 2025 study that found PFAS in 95% of 23 US beers and linked the levels to municipal drinking water (ScienceDaily). Because distillation does not carry PFAS over — they are not volatile at water’s boiling point — the plausible route into a spirit is the water added after distillation rather than the spirit itself.

Does distillation remove PFAS from water? Yes, and for the same reason boiling does not: PFAS remain behind when water evaporates. A closed still that condenses the vapour separately therefore produces PFAS-free water, whereas boiling in an open pot concentrates them (MDH).

Does reverse osmosis remove PFAS? Largely, yes. EPA reports that high-pressure membranes — nanofiltration and reverse osmosis — are “typically more than 90 percent effective at removing a wide range of PFAS, including shorter chain PFAS,” and names both as Best Available Technologies (US EPA).

What is the legal limit for PFAS in drinking water? It depends where you are. The US sets 4.0 ppt each for PFOA and PFOS (US EPA); the EU sets 0.1 µg/L for a sum of 20 PFAS and 0.5 µg/L for PFAS Total, monitored from 12 January 2026 (European Commission); WHO’s provisional values are 100 ng/L each for PFOA and PFOS. India’s IS 10500 : 2012 sets no PFAS value at all.

Is India regulating PFAS in water? Not through a parametric value yet. IS 10500 : 2012 contains no PFAS limit (IS 10500 : 2012), though PFAS have been measured in Indian surface and ground water — 15 compounds frequently detected across the Ganges basin, with PFBS up to 10.2 ng/L in river water (Sharma et al.). Producers bottling for export are generally working to the destination market’s rules.

Will a carbon filter jug remove PFAS from tap water before I add it to a dram? Not reliably. Carbon can adsorb PFAS, but performance depends on bed depth, contact time and how saturated the medium is — none of which a jug controls. Minnesota’s health department advises looking for NSF/ANSI 53 or 58 certification, while noting that current certifications do not yet demonstrate removal down to the levels EPA has set (MDH).

How much water is in a bottle of whisky anyway? For a spirit reduced from 63.5% ABV to 40%, the added water is roughly 37% of the finished bottle by volume. That share is why a contaminant in the blending water arrives in the bottle at close to a third of its concentration in the water, and why the water specification is the control point worth arguing about.

Blenders International · Water Science series