Model doubles estimated PFAS levels in US drinking water

By Daniel Okafor

A national model estimates that average PFAS concentration in US community drinking water could be 5.2 nanograms per liter for six regulated chemicals, about double the 2.6 nanograms per liter reported when below-limit results count as zero. A study published Oct. 23 in the Cell Press journal Cell Reports Sustainability produced the estimate.

What did the study measure?

The team at Texas Christian University in Fort Worth analyzed more than 1.8 million PFAS measurement records from the EPA’s near-complete January 2026 release of national drinking-water monitoring data. The dataset covered 9,991 community water systems, and the researchers modeled 25 PFAS that appeared at least once.

First author Gehendra Kharel said the project set out to determine how treating results below minimum reporting levels as zero changes the national picture of PFAS occurrence. His team used computer models to fill in values that standard summaries omit, rather than collecting new measurements.

The models generated 500 estimates of what each below-limit value could be, then averaged those estimates into a best prediction. The team applied that process to all 25 PFAS detected in the EPA dataset.

Takeaway: The team modeled 1.8 million records, including the 98% found below reporting limits.

Why does the reporting limit matter?

Under the Safe Drinking Water Act, the EPA requires public water systems to monitor 29 PFAS in drinking water. The agency also has set legally enforceable limits for six compounds, including perfluorooctanoic acid, or PFOA, and perfluorooctanesulfonic acid, or PFOS.

PFAS are man-made, so-called forever chemicals that can persist in the environment for long periods. Previous research has linked exposure to many health problems, including cancer, liver damage and developmental problems in children.

A minimum reporting level generally marks a concentration that laboratories can reliably measure and report. It does not mean a sample contains no PFAS or that the water is safe, Kharel said. Yet agencies and researchers often treat the “<MRL” notation as a measured zero when calculating averages.

Kharel put the distinction plainly: “When a water quality report says ‘not detected,’ that means below the reporting limit, not zero.” His team’s method estimates the concentration behind those entries so summaries can acknowledge uncertainty instead of discarding it.

Takeaway: A “not detected” finding records a laboratory threshold, not an absence of PFAS.

How large is the reporting gap?

The team found that 98% of PFAS test results in the EPA dataset sat below the relevant reporting limit. When researchers treated those results as zero, the summaries captured only 44% of the modeled average concentration for the 25 compounds across 9,991 systems. In other words, typical reporting omitted more than half of the estimated total.

For the six PFAS covered by enforceable EPA limits, the zero treatment produced a national average of about 2.6 nanograms per liter. The model estimated 5.2 nanograms per liter, roughly twice that amount.

A separate Environmental Working Group analysis suggests that about 176 million Americans receive water from systems where at least one PFAS was detected. That figure does not mean every sample contained a detectable PFAS, because most individual results remained below their reporting thresholds.

Takeaway: For the 25 modeled compounds, zero-based reporting accounts for 44% of the estimated average concentration.

Does the model prove a health risk?

No. Values below reporting limits remain model-generated predictions, not measurements of actual concentrations. The authors do not treat them as evidence that every below-limit result poses a health risk. The study addresses how monitoring records get summarized rather than setting new health thresholds or declaring water unsafe.

Kharel said: “Our study only examines how PFAS measurements are summarized. It does not establish that every below-limit result poses a health risk.” The authors also say their estimates could give agencies and utilities a fuller picture of PFAS levels and show where additional testing would help.

That distinction keeps the study’s scope narrow. The model can reveal where recording below-limit results as zero changes an average; it cannot replace laboratory testing or establish whether an individual sample is safe. The estimated concentrations describe possible occurrence, while health decisions require measured values and separate risk analysis.

Takeaway: The model improves interpretation of occurrence data without proving new health effects.

What has to happen next?

The method gives agencies and utilities a way to compare reported concentrations with modeled values, identify where recording below-limit results as zero changes averages, and direct additional testing where it matters most. It can also support clearer reporting of what laboratories know and what remains below their reporting thresholds.

With enforceable limits already set for six PFAS, the next regulatory milestone is whether regulators and utilities incorporate modeled values into occurrence summaries, order more testing below reporting limits and report uncertainty more clearly. Continued conversion of every “<MRL” result to zero would keep national summaries aligned with the lower 2.6-nanograms-per-liter average rather than the model’s 5.2.

Takeaway: Regulators and utilities must decide whether to use the model and add testing; that choice will determine whether the 2.6-to-5.2-nanograms-per-liter gap narrows.