PFAS in drinking water: what the new limit values measure and what they don't

The parameter ‘sum of PFAS-20’ comes from the EU Drinking Water Directive 2020/2184. Germany introduced the second, stricter value in addition. According to the Umweltbundesamt (German Environment Agency), the 4 substances it covers account for around 50% of the PFAS taken in through food. They make up around 90% of the internal body burden.

Both limit values require more than a total figure. The ordinance stipulates that the concentrations of the individual substances taken into account must be reported separately. A complete test report therefore contains the individual values, not just the sum.

One rule of the Drinking Water Ordinance becomes invisible in the limit value itself. Measured values of an individual substance below the limit of quantification are not included in the sum.

For this, Annex 7 Part I specifies a limit of quantification of 0.000 001 5 mg/l or lower as required. That is 1.5 nanograms per litre per substance. With 20 substances covered, this arithmetically amounts to just under 30 nanograms per litre that formally enter the sum with a result of zero. The limit value is 100.

In practice, the effect is smaller because laboratories generally measure more precisely. For the standard method in place since 2024, the Umweltbundesamt cites limits of quantification of 0.2 nanograms per litre in the best case. Under robust conditions, they average below 1.5. Anyone assessing an analysis report should therefore ask about the limit of quantification achieved and not just look at the sum.

The figures are expressly not representative in the statistical sense. However, they reflect the best available state of knowledge. A reliable overall picture will only come with the reporting obligation from 2026 and 2028.

Both readings of the table are correct. In about half of the samples, nothing at all is detectable. At the same time, 6.4% of the samples exceed the stricter of the two values, i.e. roughly one in sixteen. And the limit value for the sum of PFAS-4 is already ten times the toxicologically derived value.

According to the Umweltbundesamt's assessment, 4 effects are considered sufficiently supported. Established effects are a reduced antibody response to vaccinations, an increased susceptibility to infection and elevated serum cholesterol levels. In addition, there is a possible causal link with birth weight. EFSA's tolerable weekly intake is based on the first of these endpoints.

The assessment is different for other frequently cited effects. For neurological development, thyroid function and cancer risk, the Umweltbundesamt sees only insufficient evidence in the epidemiological studies. For cancer, it explicitly notes ‘no evidence’ or ‘limited evidence’. The picture is different in animal studies. There, PFOS and PFOA act as tumour promoters in the rodent liver, and PFOA can trigger Leydig cell tumours in rats.

The distinction is important because it often disappears in guide articles. An effect in rodents at high doses remains an animal finding as long as it is unresolved in humans under everyday conditions. The cancer question is subject to a separate assessment, which the following section covers.

The reasoning resolves the apparent contradiction with the Umweltbundesamt's assessment. The IARC, too, describes the evidence in humans for PFOA as limited. It restricts it to 2 types of cancer, renal cell carcinoma and testicular cancer. What counted for Group 1 was sufficient evidence in animal studies. Added to this was strong evidence on the mechanism of action in exposed humans, specifically epigenetic changes and immunosuppression.

The second caveat comes from the IARC itself. Its classifications identify a hazard, not the level of risk. In its own words, the classification says nothing about the cancer risk at different levels of exposure. 2 substances in the same group can carry widely differing risks.

PFOA was already under review in 2014 and was classified in Group 2B at the time. The 2023 upgrade is based on new data from animal studies and on the mechanism of action, not on new studies in humans. PFOS was evaluated for the first time in 2023.

According to the same source, activated carbon filtration at the waterworks currently costs less than 0.1 euros per cubic metre. The figure applies only to central plants, not to household devices.

One point applies to all methods. The removal process produces residues that themselves contain PFAS. They have to be disposed of. So a filter does not destroy PFAS; it relocates them. Our article on filtration technologies and filter methods explains how the individual technologies work.

For one substance in the PFAS group, activated carbon fails completely. Trifluoroacetic acid (TFA) consists of only 2 carbon atoms and is therefore ultra-short-chain. The Umweltbundesamt states that TFA can be removed neither by degradation nor by activated carbon filters. Only very complex methods such as reverse osmosis are effective, it says.

TFA is not a side issue. The substance is detected throughout Europe in seas, surface waters, groundwater, drinking water and precipitation. In June 2026, the Committee for Risk Assessment (RAC) of the European Chemicals Agency (ECHA) recommended classifying it as toxic for reproduction, category 1B. TFA would thus be the first chemical with a harmonised classification under the hazard classes for persistent and mobile substances introduced in 2023.

The finding has a second side that is just as much part of the truth. The health-based guideline value for TFA in drinking water is 60 micrograms per litre. Having been reviewed in light of the new assessment, it remains valid because the most sensitive endpoint is still an effect on the liver. TFA is also not one of the 20 substances in the sum of PFAS-20. Anyone who wants to remove TFA from drinking water needs a reverse osmosis system with a tight membrane. The Umweltbundesamt also describes reverse osmosis as unsustainable because of its energy requirements and the contaminated waste water stream.

Perfluorooctanoic acid has been banned throughout Europe since 4 July 2020. Delegated Regulation (EU) 2020/784 of 8 April 2020 added PFOA, its salts and PFOA-related compounds to Annex I of the EU Regulation on persistent organic pollutants (POP Regulation). The basis for this was its listing in Annex A of the Stockholm Convention. As an unintentional trace contaminant, 0.025 mg/kg remains permissible for PFOA and its salts.

What is banned is the manufacture and placing on the market of the substance, not the coated pan in the kitchen. PFOA was used as a processing aid in the manufacture of polytetrafluoroethylene; the finished polymer must be distinguished from it. The annex also contains time-limited exemptions for applications in which the limit values were initially unattainable.

This means that individual substances are banned, not the group of substances. At the European Chemicals Agency, a procedure for a broad restriction is under way, and its status is constantly changing. For the water from the tap, this changes little for the time being: past inputs remain in the groundwater regardless of future bans.

Two routes lead to reliable figures. The first is to ask your local water supplier or the responsible public health office. From 2026, the sum of PFAS-20 must be measured as part of drinking water monitoring. The individual values must be reported separately. Some suppliers were already testing before then, partly in response to 2 appeals from the Umweltbundesamt in 2020 and 2021.

The second route is your own sample, analysed by an accredited laboratory. The standard method available since 2024 uses liquid chromatography with tandem mass spectrometry. When placing the order, the Umweltbundesamt recommends asking the testing laboratory about its performance and having it documented. This means the limit of quantification and the measurement uncertainty.

There is no third route. Test strips and home methods do not detect PFAS. The relevant concentrations are just a few nanograms per litre. How water quality can be assessed overall is covered in our article on how to measure water quality.

The cartridge thus combines both operating principles: adsorption on the activated carbon for long-chain PFAS, ionic binding on the resin for short-chain ones. The same cartridge mechanically retains fine particles such as microplastics. The same rule applies to all water filter cartridges. The limit that is reached first is decisive, i.e. the date or the number of litres.

The manufacturer does not publish a retention rate in % for individual PFAS. Anyone who needs a reliable effectiveness figure for a particular water supply cannot avoid an analysis before and after the filter. Our article Why filter water explains why an additional filter stage after public water treatment can make sense at all.