Our health benchmark for PFAS: 0.5 ng/L

On every city page we compare the water with a benchmark of 0.5 ng/L (nanograms per litre) for the sum of the four most harmful PFAS: PFOA, PFOS, PFNA and PFHxS ("PFAS-4"). This benchmark is a proposal by this site, built from the published research. It is not an official limit. For the story version, read the blog post on how we got to 0.5 ng/L.

The proposal

At most 0.5 ng/L for the sum of PFOA + PFOS + PFNA + PFHxS, as an annual average.
No single one of these four substances above 0.25 ng/L.
For PFOA and PFOS, the long-term health goal remains zero.

Ideally the comparison would weight each substance by its toxicity (the Dutch RIVM's "PFOA equivalents" method). Public data does not yet allow that everywhere, so for now this site uses the simple sum of the four substances.

Why 0.5 ng/L?

In 2020, the European food safety authority (EFSA) worked out how much of these four PFAS a person can take in over a lifetime, from food and drink combined, without expected harm: 4.4 nanograms per kilogram of body weight per week, which is 0.63 ng per kilogram per day. For a 70 kg adult that is a total budget of about 44 ng per day from everything they eat and drink. The main evidence behind the number: children with more PFAS in their blood respond less well to routine vaccinations.

Sweden, Denmark and the Netherlands all built their national drinking-water limits on this same EFSA budget. So did we. Turning a "safe daily amount per person" into a "safe concentration in water" takes exactly two decisions:

Who do you protect? An adult drinks about 2 litres a day, roughly 0.03 litres per kilogram of body weight. A bottle-fed baby drinks about 0.15 litres per kilogram, five times as much for its size.
How much of the budget may water use? Food already contains PFAS, so water can only be given a share of the budget. The World Health Organization's default share for drinking water is 20%; the stricter, older convention is 10%.

Same starting number, different choices

Each authority made those two choices differently, and that alone explains why the limits differ:

Limit	Set by	Person it protects	Water's share of the budget
4.4 ng/L	Netherlands (RIVM guideline)	Adult, 70 kg, 2 L/day	20%
4 ng/L	Sweden (legal limit since 2026)	Adult	20%
2 ng/L	Denmark (legal requirement since 2021)	Small child	10%
0.5 ng/L	This site (proposal)	Bottle-fed infant	10 to 15%

None of these numbers is "what the science says". They are the same science with different safety margins: the more vulnerable the person you protect, and the smaller the share you give to water, the lower the limit. Our benchmark sits at the cautious end of the same method the three national authorities used.

For comparison, the EU-wide legal limit is far looser: 100 ng/L for the sum of 20 PFAS and 500 ng/L for "PFAS total", values negotiated without taking EFSA's 2020 assessment into account. The US enforces 4 ng/L each for PFOA and PFOS while stating that its health-based goal for both is zero. The Dutch guideline additionally weights each substance by its toxicity ("PFOA equivalents"), which is scientifically sounder than a simple sum.

Why we chose the cautious end

Food already fills most of the budget. Sweden's food agency (Livsmedelsverket) estimated that a typical adult already gets about half of the tolerable amount from food alone, and that 43% of 4-year-olds exceed the whole budget from food alone. Denmark's food institute (DTU) limited water to 10% of the budget for exactly this reason.
Sweden's own scientists reach similar numbers for children. The scientific report behind Sweden's 4 ng/L limit calculates that, after typical food intake, a 4-year-old only has room left for about 0.3 ng/L in water, and about 0.7 ng/L when water is given a 10% share. Our 0.5 ng/L sits between those two figures.
EFSA's budget itself may not be cautious enough. A 2023 reanalysis of the vaccine data (details below) points to considerably lower safe levels, the US health-based goal for PFOA and PFOS is zero, and PFOA is classified as carcinogenic to humans (IARC Group 1).

The arithmetic, so you can check it

How much of EFSA's daily budget the water alone would use up:

Water at	Adult: 70 kg, drinks 2 L/day	Bottle-fed infant: 0.15 L per kg per day
4 ng/L (Sweden's current limit)	18% of the budget	95% of the budget
1 ng/L	4.5%	24%
0.5 ng/L (our benchmark)	2.3%	12%

Our benchmark comes straight from the infant column: 0.63 ng per kg per day, times a 10 to 15% share for water, divided by 0.15 litres per kg per day, gives 0.4 to 0.6 ng/L. We rounded to 0.5 ng/L.

One honest caveat: EFSA designed its budget for lifetime intake, not literally for babies, whose exposure through pregnancy and breastfeeding is modelled separately. We use the bottle-fed infant as the most-exposed realistic person, to show how little room 4 ng/L leaves, not as a strict infant risk model. Even read that way, the table shows the problem with 4 ng/L: a highly water-exposed infant could use up nearly the whole budget from water alone, before any food, placental transfer or breast milk is counted.

Is literally any PFAS too much?

Not in the simplistic sense that one molecule causes meaningful harm. "PFAS" covers thousands of chemically different substances; a short-chain PFAS that the body clears quickly cannot be treated like PFHxS, which can stay in human blood for years. Estimated serum half-lives are roughly 5.3 years for PFHxS, 3.4 years for PFOS and 2.7 years for PFOA.

But for persistent, bioaccumulative PFAS:

there is no biological benefit from exposure;
exposure accumulates over years;
epidemiology has not reliably identified a reassuring low-effect threshold;
PFOA is classified as carcinogenic to humans (IARC Group 1), and PFOS as possibly carcinogenic (Group 2B).

A 2023 reanalysis of vaccine-response data obtained serum benchmark levels around 1.5 to 4 ng/mL, well below the 17.5 ng/mL EFSA selected. That analysis is not yet regulatory consensus, but it suggests EFSA's tolerable intake is not especially conservative.

So "zero" is the right directional goal, and 0.5 ng/L is our practical benchmark. Concentrations around 1 to 4 ng/L do not create an acute danger or guarantee disease; they are an avoidable contribution to the body burden that builds up over a lifetime. We treat 4 ng/L as an action threshold, not an ideal safety target.

How this site applies the benchmark

We add up the four substances from the latest public check; the benchmark itself is meant as an annual average.
Substances too small to measure count as zero, so totals are lower bounds. When the lab's detection limits are too coarse to conclude anything, the page says so instead of showing a false pass.
The benchmark is stricter than every current legal limit: water "above our benchmark" can still be perfectly legal.

How this page was made

The research synthesis and the exposure calculations were prepared with the help of AI, then checked against the primary sources cited below. The benchmark is a documented proposal, not an official standard and not medical advice. If you are worried about your water, contact your supplier or your local health authority.

Sources

This site is a data project, not an official source. If in doubt, check with your water supplier or local authority.