"Insufficient PFAS Safety Data": Did the FDA’s Cosmetics Report Ask the Wrong Question?
Counts, use levels, and wastewater loading all tell different stories
The FDA recently did something unusually candid for a federal consumer-products conversation. It released a congressionally mandated assessment of PFAS in cosmetics and, in plain language, said it cannot determine safety for most of what it found because the underlying data are not there.
That is not a trivial headline, as it signals that we are building policy, litigation narratives, and corporate reformulation timelines on an evidence base that is still fragmented in three places: 1) how we identify PFAS in formulations, 2) how we translate ingredient presence into real exposure, and 3) how we handle what PFAS become after use, disposal, and treatment. If we do not separate those problems, “PFAS in cosmetics” turns into an argument about confidence rather than a pathway to better decisions.
So, what the FDA report actually says, and what it does not.
Start with the concrete, because this report is not purely conceptual. Using mandatory cosmetic product listing data, the FDA cross-referenced an inventory of 177 PFAS and found that 51 PFAS appeared in 1,744 cosmetic formulations as of August 30, 2024, which it reports as 0.41% of registered cosmetic products (430,134). The report also flags why that percentage should not be read as a stable truth. In other words, listings can change over time, exemptions exist (including small business exemptions), and underreporting remains possible.
The FDA then prioritized the top 25 PFAS by frequency of use and attempted a safety evaluation. The outcome is the part that deserves more attention than the “0.41%” figure. The report concludes that for 19 of the 25 PFAS, it cannot reach a safety conclusion due to lack of critical toxicological data. Five were judged to pose low safety concern under intended use conditions, and one, perfluorohexylethyl triethoxysilane, was flagged as a potential concern at the highest assumed use level in body lotion, with the report explicitly walking through uncertainty drivers such as unknown use levels, lack of dermal absorption data, and reliance on limited unpublished study information.
Equally important is what the FDA press announcement says the report is focusing on: intentionally added PFAS as ingredients, not PFAS present as contaminants. That scope choice may be defensible for a first pass, but it creates an immediate interpretive trap. In practice, “intentional use” is not the only way PFAS enter a product. Residuals, processing aids, polymerization byproducts, and impurities can drive exposure and emissions even when the ingredient list looks clean. If stakeholders treat “intentionally added” as synonymous with “all PFAS that matter,” we will be back in the same argument in a year, just with better spreadsheets.
Now to the identity problem & why “search for fluoro” fails in the real world.
One of the most revealing lines in the FDA report is not a risk conclusion. It is a correction. The report notes that polyvinylidene fluoride was not retrieved in their search but was confirmed as a PFAS in the wINCI database and had to be added to the inventory, bringing the total to 177 PFAS.
That is the quiet admission behind a lot of public PFAS debates. Name-based screening is brittle. It misses PFAS without obvious “fluoro” strings, it misses polymers and crosspolymers whose nomenclature does not advertise the fluorinated moiety, and it is vulnerable to trade-name opacity. A strong inventory needs structure-based identifiers, CAS linkages where possible, and an explicit strategy for polymers and side-chain fluorinated materials, which are often exactly where analytical coverage is weakest.
This is also why third-party analytical screening studies keep landing like a surprise, even though they should not. The 2021 Environmental Science & Technology Letters paper by Whitehead and colleagues used total fluorine as a screen across a large set of products and reported that foundations, mascaras, and lip products had the highest proportion of high fluorine concentrations. Total fluorine is not PFAS speciation, and it cannot tell you which structures are present, but it is a powerful reminder that label and listing approaches can systematically under-detect fluorinated content when disclosure is incomplete or nomenclature is opaque.
Next is the exposure problem & why “0.41% of products” is not an exposure metric.
Counts are politically attractive because they are easy to communicate. They are also easy to misinterpret.
The FDA’s 0.41% figure is a fraction of registered products containing one of the listed PFAS ingredients. It does not reflect use concentration, frequency of consumer use, amount applied, how often a product is rinsed off, and whether the PFAS present is a low-mobility polymer or a precursor that can transform. Even within the report, the agency acknowledges material uncertainties in use levels and route-relevant data like dermal absorption.
This is where scientific evidences are useful. Balan et al. in Environmental Science & Technology (2024) tackled a different question than the FDA did. Instead of asking “how many products,” they asked “how many kilograms,” using ingredient databases, manufacturer-reported concentrations, and California sales data to estimate mass flow for a product class that is, by design, emitted to wastewater.
They estimated that cosmetics sold in California over a one-year period contained 650 to 56,000 kg total PFAS, with corresponding organic fluorine and fluorinated side-chain mass ranges, and they found that over 90% of the estimated PFAS mass came from shaving creams and gels, hair care products, facial cleansers, sun care products, and lotions and moisturizers. All nine makeup subcategories combined accounted for under 3% of total PFAS and fluorine mass in their estimate.
That is not an argument that makeup is not important. It is an argument that “importance” depends on the question. If you are doing dermal or inhalation exposure prioritization, makeup categories can matter disproportionately, and high-fluorine screening supports that. If you are doing wastewater source control, rinse-off categories may deliver more leverage per unit effort. Both can be true, and policy gets distorted when we pretend there is only one “PFAS in cosmetics” problem.
Another layer is the transformation problem, and why long-term safety is not just about the named ingredient.
The FDA report does discuss route-of-exposure logic, including dermal, inhalation for sprays and powders, and incidental ingestion for lip products, and it frames its assessment in standard risk assessment terms when data allow. The weak link, which the report indirectly exposes, is that the most consequential chemistry questions are often not “is ingredient X toxic,” but “what fraction of the formulation is actually ingredient X,” “what impurities ride along,” and “what does it transform into during use and after disposal.”
Cosmetics are not static chemicals in a vial. They sit on skin under UV, they contact oxidants, they get washed into chlorinated wastewater, and they move into sludges, landfills, and sometimes thermal treatment systems. When regulators focus narrowly on intentionally added ingredients, the system-level risk can migrate into the spaces that are hardest to inventory and hardest to measure, impurities and transformation products.
In fact, published data on PFAS in personal care products keeps emphasizing breadth, not just a short list of familiar acids. Harris et al. (2022) reported a breadth of PFAS in cosmetics and personal care products, including classes that have been measured in wastewater and human blood, highlighting both human and environmental exposure relevance. The details differ by product type, and the analytic challenges are real, but the direction is consistent: formulation complexity and PFAS diversity create blind spots if we rely on a single inventory method.
Why this matters now, and what should change next?
The regulatory context is accelerating, mostly outside Washington. California’s AB 2771 prohibits, since January 1, 2025, the manufacture and sale of cosmetics with intentionally added PFAS. Other states have moved on similar timelines, and the compliance burden is landing on companies that still lack consistent definitions, validated methods, and supply-chain transparency. Internationally, Europe is actively evaluating broad PFAS restrictions under REACH, with cosmetics explicitly within scope discussions. Meanwhile, OECD has already published work specifically on PFAS and alternatives in cosmetics, which is a quiet indicator of where regulators and markets expect the conversation to go, substitution, performance, and feasibility, not just detection.
So the constructive path forward is ask the practical question is how to turn “insufficient data” into a targeted research and reporting agenda that improves decisions for regulators, industry, and downstream managers.
My view is that the next iteration of this space needs to be built around three parallel upgrades. First, inventory needs to move beyond string-matching and toward structure-aware registries that handle polymers, crosspolymers, and side-chain fluorinated materials explicitly, with clear linkage between INCI naming and chemical identity. For example, the PVDF miss should be treated as a design lesson, not an anecdote.
Second, exposure science has to become route-realistic. For cosmetics, dermal absorption and inhalation from powders and sprays cannot remain “assumed” parameters, especially for the PFAS classes that dominate listings. The FDA’s uncertainty discussion for perfluorohexylethyl triethoxysilane is a case study in how quickly a risk conclusion becomes a sensitivity analysis when dermal data and use levels are missing.
Third, the sector should stop treating wastewater and solid residuals as an afterthought. If the objective is meaningful risk reduction per dollar, source control needs a mass-flow lens. Current studies translated product composition into system loading and highlight which categories plausibly drive emissions. This is also where regulators can start better practices. A ban defined around “intentionally added PFAS” is one tool, but without verification strategies that address impurities and transformation potential, the market can comply on paper while externalizing risk into harder-to-measure chemical space.
The FDA report is being read as a consumer safety headline. I read it as a governance diagnostic. If we respond by building better identity infrastructure, route-relevant toxicology, and mass-flow prioritization, the next report can be more than a reflection of uncertainty. It can be a map for decisions that hold up in court, in treatment plants, and in real products on real skin.