TL;DR — The PFAS-and-floss literature is two cross-sectional studies that disagree, and a handful of direct measurements that don't. What survives scrutiny is narrow: PTFE is by definition a PFAS, total-fluorine screening found fluorinated material in 6 of 18 floss products, and the headline serum finding — from the smaller study, on a confidence interval of 0.2–55.7 — was contradicted in direction by an analysis 38 times its size. No study has shown causation, and none has quantified a dose.
Two cross-sectional studies carry almost the entire "PFAS in dental floss" story, and they point in opposite directions. The first, published in 2019, tested 178 middle-aged women and associated use of one PTFE-based floss with a 24.9% higher serum level of PFHxS — 95% confidence interval 0.2 to 55.7. The second, published in 2025, analyzed 6,750 U.S. adults in NHANES 2009–2020 and found that floss users had lower overall serum PFAS than non-users, with PFOA the single exception. Neither study design can show cause and effect. This article walks through what each one actually measured.
The rest of this cluster covers the material side: the pillar sets out six filament materials against five fixed criteria, one comparison piece puts silk, nylon, and PLA head to head, and a ranked piece scores six material categories on certification, disposal, and cost per meter. This one stays on the evidence itself.
The two studies, side by side
| Criterion | Boronow et al., 2019 | Jiao et al., 2025 |
|---|---|---|
| Design | Cross-sectional | Cross-sectional |
| Participants | 178 middle-aged women | 6,750 adults 30+ |
| Population | African American and non-Hispanic white women | NHANES, nationally representative |
| Exposure measure | Self-reported use of a named product | Self-reported floss use, no brand or material |
| Product testing | 18 floss products screened for fluorine | None |
| Headline result | +24.9% serum PFHxS (95% CI 0.2–55.7) | Lower overall serum PFAS; PFOA slightly higher |
| Can it show causation? | No | No |
Same design, opposite direction, and a 38-fold difference in sample size. That is the whole controversy in one table.
What the 2019 study measured
The Silent Spring Institute study associated flossing with a PTFE-based product with 24.9% higher serum PFHxS in 178 women, and separately screened 18 floss products for fluorine using PIGE spectroscopy. Fluorine was detected in 6 of the 18: all three products from the PTFE-based brand tested, two of three store brands carrying "compare to" labeling, and one product that described itself as single-strand Teflon fiber.
Two separate findings, then, and they should not be merged:
- The product assay. Some floss contains fluorinated material. This is a direct measurement, and it is the strongest part of the paper.
- The serum association. Women who reported using one particular product had higher PFHxS. This is an epidemiological association from a questionnaire.
The institute's own press release framed it as "associated with a higher body burden" — associated, not caused. That distinction was lost in most of the coverage that followed.
Does total-fluorine screening prove anything on its own?
Partly. Measuring total fluorine is a screening method: it tells you fluorinated compounds are present, not which ones. The method has been validated elsewhere — in a cosmetics study by the same laboratory group, 29 samples sent for targeted analysis after fluorine screening each contained between 4 and 13 specific PFAS. So a positive fluorine screen is a reasonable first step. It is not identification.
The confidence interval problem
The reported effect is 24.9%, and the 95% confidence interval runs from 0.2 to 55.7. Read that interval literally: the data are compatible with an increase of roughly a fifth of one percent, and equally compatible with an increase of more than half. The result clears the conventional significance threshold by a margin so thin that it would flip with a handful of reclassified participants.
Figure 1 — One result, three numbers: the 2019 PFHxS effect and its interval
Wide intervals are what small samples produce. With n=178 and a subgroup of self-reported users of one product, that is the expected result, not a flaw the authors hid. It does mean the point estimate of 24.9% should never be quoted without the interval attached.
The chemistry objection
The sharpest technical criticism came from the American Dental Association, and it is not a rhetorical one. The study used fluorine as a marker for PTFE in the floss — but the compound elevated in the women's blood was PFHxS, and the ADA argued that detecting fluorine in a product "does not mean that it is the source of the PFHxS in the women". PTFE is a polymer; PFHxS is a short-chain perfluoroalkyl sulfonate. The link between the two is a manufacturing-residue hypothesis, not a demonstrated pathway.
A trade association has an obvious interest in not scaring people away from flossing. The chemical objection stands on its own regardless. It was echoed from a non-commercial source: a McGill University science-communication piece noted that there is no way to know the compounds in the blood came from the floss rather than from any of the dozens of other documented exposure routes.
What the 2025 NHANES analysis found
The largest attempt to reproduce the finding used NHANES 2009–2020, 6,750 adults aged 30 and over. It reports that people who used dental floss had lower serum PFAS overall, with the exception of PFOA, which was slightly elevated. The absolute gap on PFOA is small: 2.534 versus 2.605 ng/mL between groups, a difference of roughly 0.07 ng/mL.
Its limitations are as real as the 2019 study's, and different in kind:
- No brand, no material. NHANES asks whether you floss. It does not ask what the floss is made of. A PTFE effect, if one exists, would be diluted across every material in the market.
- Confounding runs the wrong way. People who floss daily differ systematically from people who never do — income, education, diet, water source. Any of those correlate with PFAS exposure independently.
- Still cross-sectional. Blood was drawn once. Exposure was recalled once. No sequence, no follow-up.
The 2025 paper concludes that more research is needed to clarify the role of floss in PFAS exposure. That is the honest reading of both papers combined.
Evidence ranked, strongest claim to weakest
Applying five criteria — design, sample size, exposure measurement quality, effect precision, and independent replication — the claims in this area sort out roughly as follows.
Figure 2 — Six claims, scored on evidence type, replication, and strength
1. PTFE is a PFAS. Not contested. A peer-reviewed lifecycle review concluded that there is no scientific rationale for treating fluoropolymers as low-concern and separate from other PFAS. This is a classification question, settled by definition and chemistry rather than by epidemiology. Note the scope: the review addresses the full lifecycle including manufacture and disposal, not a consumer's exposure from finished floss in the mouth.
2. Some floss products contain fluorinated material. Direct measurement, 6 of 18 products, one laboratory, one method. Strong for what it covers; the sample is small and dated.
3. PFAS persist in the body once absorbed. In a cohort of 26 airport workers exposed via drinking water, of whom 17 were followed monthly, measured serum half-lives were 1.77 years for PFOA, 2.87 years for PFHxS, and 2.93 years for linear PFOS, against 44 days for the short-chain PFBS. Small cohort, high exposure, and other studies report longer half-lives — but the direction is not in dispute.
4. Floss use is associated with higher serum PFHxS. One cross-sectional study, n=178, confidence interval nearly touching zero, contradicted in direction by a study 38 times larger. Weak.
5. Floss meaningfully contributes to an individual's PFAS body burden. Not established. No study has measured a dose, a transfer rate, or a before-and-after.
6. Floss-derived PFAS causes health effects. Not established, and not close. U.S. federal health authorities describe the PFAS health literature in terms of association, with the ATSDR stating plainly that "causal relationships have not been established for these health effects" — and that evidence comes from populations exposed at levels far above anything a floss pathway could plausibly deliver.
Regulation moved anyway
Regulators did not wait for the epidemiology to resolve, which is worth noting as context rather than as evidence. In the United States, dental floss is a Class I medical device exempt from premarket review; the regulation describes it as a string-like device of cotton or other fibers and says nothing about coating chemistry. No agency assesses the materials before sale.
At state level, one concrete prohibition already exists. Minnesota's Amara's Law banned the sale of products with intentionally added PFAS in eleven named categories from January 1, 2025 — dental floss among them. That is one state, not a federal ban.
Verdict
The defensible position is narrow. Fluorinated material has been measured in some floss products, and PTFE belongs to the PFAS class — both direct findings. The claim that flossing raises your PFAS burden rests on a single small cross-sectional study whose confidence interval barely excludes zero, and the largest attempt to reproduce it found the opposite overall direction. No study has established causation, and none has quantified a dose.
Anyone weighing this should read the two studies as what they are: hypothesis-generating, not conclusive. Questions about your own oral care routine belong with your dentist, who can weigh them against your actual dental condition.
