Models Suggest US Drinking Water May Harbor Double Reported PFAS Concentrations

Conventional methods for reporting perfluoroalkyl and polyfluoroalkyl substances, widely referred to as PFAS or forever chemicals, could underestimate the average concentration of these synthetic compounds in United States drinking water by as much as half, according to a computational modeling study published in the journal Cell Reports Sustainability. Researchers from Texas Christian University in Fort Worth determined that widespread monitoring assessments regularly treat chemical concentrations falling below analytical minimum reporting levels as zero. By relying on this zero-substitution convention, current summaries capture only about 44 percent of the average concentration across 25 distinct PFAS compounds evaluated in community water systems nationwide, leaving more than half of the model-estimated chemical presence unaccounted for in standard reporting summaries.

Under the regulatory framework of the Safe Drinking Water Act, the United States Environmental Protection Agency mandates that public water systems monitor 29 separate PFAS compounds in drinking water supplies. These manufactured substances are known to persist in natural environments over prolonged periods, and prior scientific investigations have established associations between PFAS exposure and serious human health complications, including various forms of cancer, liver damage, and developmental disorders in children. In January 2026, the Environmental Protection Agency released an extensive, near-complete monitoring dataset covering water utilities across the United States. An analysis of these monitoring records conducted by the Environmental Working Group indicated that approximately 176 million people across the nation receive tap water from utilities where at least one PFAS compound was detected.

Despite this widespread detection across municipal supplies, the vast majority of individual test records within the federal dataset fail to reach the threshold that laboratories can routinely quantify. As a result, the overwhelming majority of individual entries are recorded with the designation of falling below the minimum reporting level, or less than MRL. Lead study author Gehendra Kharel, an environmental researcher at Texas Christian University, explained that a minimum reporting level represents the baseline concentration that most analytical facilities can reliably quantify and report under standard operating procedures. Crucially, Kharel noted that an entry falling below this detection boundary does not indicate that tested water is safe or entirely free of PFAS. Nevertheless, when regulatory agencies and academic researchers calculate how widespread these contaminants are or compile national summary averages, these below-threshold records are routinely treated as numerical zeros.

To investigate the mathematical implications of treating sub-threshold measurements as zero, Kharel and colleagues conducted a comprehensive analysis of more than 1.8 million individual PFAS measurement entries contained in the nationwide Environmental Protection Agency database. Their initial evaluation established that 98 percent of all recorded PFAS test results in the national collection fell below the agency’s established minimum reporting levels. Seeking to understand how treating those entries as zero alters the overall understanding of contaminant occurrence, the researchers developed computational simulation models to estimate possible actual concentrations for records labeled as below the reporting limit. For each below-limit record, the computer model generated 500 distinct potential concentration values, which the investigative team subsequently averaged to establish an optimal statistical prediction. This simulation was performed systematically for all 25 PFAS compounds that had been detected at least once within the agency’s monitoring data.

The resulting computational estimates demonstrated that the conventional practice of treating sub-limit readings as zero captures merely 44 percent of the aggregate mean concentration of the 25 detected PFAS across 9,991 community water systems evaluated nationwide. This systematic underestimation is also apparent among the specific chemical substances subject to federal standards. The Environmental Protection Agency has established legally enforceable drinking water limits covering six particular PFAS compounds, including perfluorooctanoic acid, known as PFOA, and perfluorooctanesulfonic acid, known as PFOS, both of which have been linked to severe health conditions. When monitoring analyses treat below-limit values as zero, the calculated national average concentration for these six regulated substances is approximately 2.6 nanograms per liter. However, the simulation model developed by Kharel’s team estimates that the true average concentration across drinking water systems could in reality reach 5.2 nanograms per liter, representing a doubling of conventional estimates.

The study authors explicitly outlined the structural scope and limitations of their computational findings, distinguishing model predictions from empirical laboratory measurements. Kharel emphasized that the investigation focuses specifically on how environmental measurement datasets are mathematically aggregated and summarized, rather than asserting that every below-limit chemical concentration carries an immediate health hazard. The authors noted that the published values represent computer-generated predictive estimates rather than direct physical quantifications of contaminants existing beneath analytical detection thresholds. Seed funding for the research project was provided by the Natural Resource Research Initiative at Texas Christian University, supporting the team’s effort to quantify how data censoring influences public monitoring conclusions without overstating the biological risks of individual low-level exposures.

Despite these operational caveats, the researchers concluded that their censoring-aware modeling framework provides a valuable approach to constructing a more complete understanding of chemical occurrence in American drinking water and refining the interpretation of environmental monitoring data. Kharel pointed out that when consumer water quality reports list a chemical contaminant as not detected, the designation signifies only that the compound is present below the laboratory’s reporting limit rather than completely absent. By accounting for the difference between a sub-threshold detection limit and a true zero value, the modeling methodology provides water utilities and environmental oversight agencies with a practical tool to recognize where expanded testing could be beneficial and to communicate monitoring uncertainties with greater scientific clarity.

Subject of Research: PFAS concentrations and reporting limits in United States drinking water

Article Title: US drinking water could contain double the reported PFAS levels, model estimates

Article References: US drinking water could contain double the reported PFAS levels, model estimates. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: PFAS, drinking water, water quality, environmental health, Safe Drinking Water Act, computational modeling, Environmental Protection Agency, public water systems, drinking, water, contain, double