Does Water Fluoridation at Recommended Levels Lower Children's IQ?

Precise question and why it matters

Community water fluoridation (CWF) is one of the most widely deployed public-health interventions of the past eighty years, and it is now being dismantled on the basis of a claim that it damages children's brains. This brief asks a deliberately narrow question: does exposure to fluoride at the concentration used in CWF — about 0.7 mg/L in the United States, and no higher than 1.5 mg/L under the World Health Organization's drinking-water guideline — reduce children's IQ, and by how much?

The stakes are large and the policy is already moving. In September 2024 a U.S. federal district court held, under the Toxic Substances Control Act, that water fluoridation at 0.7 mg/L presented "an unreasonable risk of injury to health," and ordered the Environmental Protection Agency to begin rulemaking; the EPA appealed in January 2025. In April 2025 the U.S. health secretary directed the CDC to stop recommending fluoridation, and states including Utah and Florida moved to restrict it. Roughly two-thirds of the U.S. population and about 400 million people worldwide drink fluoridated water. The decisions turn on whether a real finding at high exposure travels down to the exposure the intervention actually delivers.

I use four labels throughout. An observation is an association measured in a single study. A quasi-causal estimate comes from a design that exploits plausibly exogenous variation — a natural experiment. A causal estimate comes from randomization, which for this question has never been done and almost certainly never will be. An extrapolation carries a result from one exposure range, population, or design into another.

Best current answer

Three conclusions are well supported, and one claim is not:

• The association at high exposure is real. Above roughly 1.5 mg/L — mostly naturally occurring groundwater fluoride in China, India, Iran, Pakistan, and Mexico — more fluoride is consistently associated with lower measured IQ. This is a robust, dose-dependent observational pattern (~90%).
• A detectable IQ reduction at 0.7 mg/L is not established. The best-identified studies at policy-relevant exposure — a Swedish registry natural experiment, a U.S. nationally representative cohort, and a U.S. state-representative cohort followed to age 80 — find no harm, and two find a small benefit. The meta-analysis that launched the policy shift found no water-fluoride effect below 1.5 mg/L when it pooled all studies.
• The dental benefit is modest in modern data, and the trade-off is real. The 2024 Cochrane update finds about a quarter of a tooth less decay in primary teeth, with low certainty, and three cessation studies find more caries after fluoridation stopped. Any honest assessment compares two small, uncertain effects; it is not a verdict on a single outcome.
• The claimed low-dose harm is an extrapolation, not a finding. The 0.7 mg/L intervention sits below the range where the signal lives, and no mechanism has been demonstrated in humans at that range.

Three exposures share one name

Much of the public dispute is a category error. "Fluoride exposure" refers to at least three different quantities that are not interchangeable:

• Water concentration at the source — the policy variable, 0.7 mg/L.
• Urinary fluoride — a biomarker of recent total intake from all sources, including water, black tea, dental products, and food.
• Estimated total intake — a modeled dose.

A community's water concentration is not an individual's urine concentration. The National Toxicology Program (NTP) notes that total exposure varies widely even among people in optimally fluoridated areas, because intake depends on how much water and tea a person drinks and how much fluoride toothpaste they swallow. The distinction matters because the pooled group-level water estimate in the largest meta-analysis (SMD −0.45) is three times the individual-level dose-response estimate (SMD −0.15) — the classic signature of ecological data, where group averages erase the individual variation that would otherwise push the estimate toward the null.

The policy question is about a water concentration. The strongest signals come from biomarkers in high-exposure regions. Bridging those two is exactly the step that is missing.

What the high-exposure evidence shows

The NTP's 2024 monograph, a systematic review of human, animal, and mechanistic evidence, concluded with moderate confidence that "higher estimated fluoride exposures (e.g. … drinking water fluoride concentrations that exceed the [WHO] 1.5 mg/L … guideline) are consistently associated with lower IQ in children." It found only low confidence for other neurodevelopmental outcomes, no evidence of an adult cognitive effect, and — importantly — judged the animal and mechanistic literature too sparse and heterogeneous to establish biological plausibility. The monograph explicitly did not address whether fluoridation at 0.7 mg/L has any measurable effect.

The quantitative companion, Taylor and colleagues in JAMA Pediatrics (2025), pooled 74 studies: 64 cross-sectional and only 10 prospective cohorts, with 45 conducted in China and none in the United States. Fifty-two of the 74 were rated high risk of bias. Its results:

The individual-level prenatal cohorts are the most credible part of this literature and the part a careful reader should engage with, not dismiss:

• Mexico (ELEMENT; Bashash et al. 2017): among 299 mother-child pairs, a 0.5 mg/L increase in maternal urinary fluoride predicted 3.15 lower GCI points (95% CI −5.42 to −0.87) and 2.50 lower IQ (−4.12 to −0.59).
• Canada (MIREC; Green et al. 2019): among 512 pairs, a 1 mg/L increase in maternal urinary fluoride was associated with 4.49 lower IQ in boys (−8.38 to −0.60) and no significant association in girls.
• Canada (MIREC; Till et al. 2020): among 398 dyads, formula-fed infants in fluoridated cities had lower Performance IQ — 9.3 points per 0.5 mg/L water fluoride (−13.77 to −4.76).
• United States (MADRES; Malin et al. 2024): among 229 predominantly Hispanic mother-child pairs in Los Angeles, a 0.68 mg/L increase in third-trimester maternal urinary fluoride was associated with nearly double the odds of a borderline-clinical Total Problems score at age 3 (OR 1.83, 1.17 to 2.86).

These are real, prospective, individual-level signals. They are also small in IQ terms, inconsistent across subgroups, and — in the Canadian data — sex-specific in a way that no mechanism explains. The question is whether they reflect fluoride or whatever travels with it.

The anomaly: the target sits below the signal

Here is the central fact that the policy debate has not absorbed. The JAMA Pediatrics meta-analysis, the load-bearing document, found that the water dose-response association was null when exposure was restricted to below 1.5 mg/L in all studies. It became inverse below 1.5 mg/L only in the subset of low-risk-of-bias studies — but that subset is small and dominated by the same prenatal cohorts above. The NTP reached moderate confidence for exposures above 1.5 mg/L. The court found an "unreasonable risk" at 0.7 mg/L and the CDC's recommendation was withdrawn, but 0.7 mg/L is less than half of 1.5 mg/L, and the judge's reasoning rested on the absence of a safety margin, which is a precautionary judgment, not a measured effect.

At the exposure the intervention actually produces, the higher-quality evidence is null. That is the anomaly: a genuine high-dose finding is being used to condemn a low-dose intervention without the bridging evidence that would justify it.

The low-exposure cohort record: null and positive

When prospective cohorts measure exposure at policy-relevant levels, the inverse signal does not appear:

Three of these deserve emphasis. The Swedish study (Aggeborn & Öhman, Journal of Political Economy) exploits exogenous geological variation in natural fluoride and finds a zero effect on cognitive ability, a positive effect on dental health, and a 4.4% income gain per 1 mg/L — concentrated among lower-socioeconomic-background individuals. The U.S. High School and Beyond analysis finds that children exposed to recommended levels performed modestly better in secondary school, an advantage that faded by age 60. The Wisconsin Longitudinal Study — the first U.S. analysis to track CWF exposure against measured adolescent IQ and cognition to age 80 — finds no negative association at any age. The Brazilian meta-analysis, which restricted inclusion to 0.4–1.5 mg/L, likewise finds no association.

If there were a large, general cognitive penalty from CWF at 0.7 mg/L, these designs should have seen it. They did not.

Why the two literatures disagree

Four explanations account for most of the gap, and they are not mutually exclusive:

1. Risk of bias. The dose-response meta-analysis (Veneri et al. 2023) found the inverse association was "particularly strong in the studies at high risk of bias, while no adverse effect emerged in the only study judged at low risk of bias." The effect is concentrated where the methods are weakest.
2. Confounding, especially by place and class. High-fluoride areas are disproportionately rural, poorer, and exposed to other neurotoxicants — arsenic, lead, manganese — with different nutrition and different IQ-testing norms. Broadbent and colleagues noted explicitly that the high-exposure associations "may have been affected by confounding, particularly by urban or rural status." None of the Chinese cross-sectional studies can rule this out.
3. Ecological inflation. Pooling group-level rather than individual-level data exaggerates the apparent effect (SMD −0.45 versus −0.15), and the largest pooled estimate carries severe heterogeneity (I² ≈ 94%) and evidence of publication bias.
4. Extrapolation. The clearest studies of the intervention point the other way, and no mechanism has been demonstrated. The NTP itself judged the mechanistic and animal evidence inadequate; a 2026 narrative review in the Journal of the American Dental Association, two of whose three authors sit on the American Dental Association's National Fluoridation Advisory Committee, states flatly that "there is no established biological mechanism to explain [fluoride's] effect on IQ."

None of this proves fluoride is harmless at every dose. It shows that the harm claim, at the dose in question, rests on the observational end of the evidence while the better-identified end is null.

The benchmark-concentration claim

The strongest version of the low-dose case comes from Grandjean and colleagues (2024), who merged the Danish Odense cohort with the Mexican and Canadian cohorts. The Danish cohort alone showed no significant association (β = 0.08 per doubling of exposure). Merged, the three studies yielded a joint benchmark concentration of 0.45 mg/L with a lower confidence limit of 0.28 mg/L, which the authors describe as "an approximate threshold for developmental neurotoxicity." A separate dose-response meta-analysis reported a weaker, non-significant decline of 2.15 IQ points in urinary fluoride above 0.28 mg/L.

The problem is the inferential step. A benchmark concentration is a statistical quantity derived from a dose-response fit; its lower confidence limit being below a water concentration does not demonstrate that a real effect exists there, especially when the individual cohorts were individually null or inconsistent. Reporting a BMCL as a "threshold" converts a modeling output into a safety claim. The NTP declined to make that conversion, and the meta-analytic confidence intervals at low dose include the null.

The dental trade-off

The other side of the ledger is smaller than it once looked. The 2024 update of the Cochrane review of water fluoridation included 157 studies, all non-randomised, and separated data collected before and after fluoride toothpaste became widespread in 1975. On contemporary evidence, starting fluoridation may reduce decayed, missing, and filled primary teeth (dmft) by about a quarter of a tooth per child (mean difference 0.24; 95% CI −0.03 to 0.52; two studies, 2,908 children) and may raise the share of caries-free children by about 4 percentage points in primary teeth and 3 in permanent teeth. Each of these estimates is rated low certainty, and each interval includes no benefit. The review was very uncertain about decay in permanent teeth and could not size the effect of stopping fluoridation from its single eligible cessation study. The larger figures from its 2015 version, a 35% reduction in dmft and 26% in DMFT, rested mostly on studies from before 1975 and no longer describe the modern increment.

The cessation studies outside Cochrane's inclusion criteria point the same way, with wider gaps. Calgary, which stopped fluoridating in 2011, had 64.8% primary-dentition caries prevalence by 2018/2019 versus 55.1% in still-fluoridated Edmonton, a gap that widened over time and survived adjustment for confounders. Juneau, Alaska, saw more caries procedures and 28–111% higher caries-related treatment costs five years after cessation. In Alberta, caries-related dental treatments under general anesthesia rose in the non-fluoridated city after cessation, most sharply in children under 6. These are observational comparisons between cities, so they cannot rule out other local changes, but three settings agree on the direction, and two of them measure costs that fall on children with the least access to care: Medicaid-eligible children in Juneau and young children treated under general anesthesia in Alberta.

This is the balance the cognitive literature should be read against. Even if the small inverse associations in the prenatal cohorts were partly causal, they are on the order of one to a few IQ points at the most extreme exposures, well above 0.7 mg/L. At recommended levels the comparison is between a small, low-certainty dental benefit and a cognitive harm that the best-identified studies do not detect. That makes the policy judgment closer than either side usually presents it, and more dependent on local caries risk and access to dental care.

Evidence quality and limitations

Strengths. The discussion now includes well-identified designs the earlier literature lacked: a registry natural experiment with exogenous exposure (Sweden), two population-representative U.S. cohorts, and policy-driven before-after comparisons for caries. The two largest syntheses registered their methods and used formal risk-of-bias tools. The individual-level prenatal cohorts measured exposure with biomarkers rather than group averages.

Weaknesses. The meta-analytic core is cross-sectional, geographically clustered, and mostly high risk of bias; only 10 of 74 studies were prospective. No randomized trial has ever assigned fluoridation and measured cognition, and none plausibly will. Urinary fluoride is an imperfect biomarker of brain-relevant exposure, and the biomarker-versus-water question is unresolved. The U.S. studies infer exposure from historical water records rather than individual intake.

What I did not do. No systematic search protocol was run; the base came from targeted Crossref, Europe PMC, and web queries, so a relevant study could have been missed. Every effect size quoted here was read from the source abstract or full text, not from a secondary summary, but no de novo analysis was performed.

Calibrated confidence

• ~90% that higher fluoride exposure (above ~1.5 mg/L) is associated with lower children's IQ. Basis: consistent direction across dozens of studies, a dose-response gradient, and the NTP's moderate-confidence rating with independent replication.
• ~55% that this high-exposure association is partly causal rather than entirely confounded. Basis: plausible dose-response and some prospective support, held below certainty by pervasive risk-of-bias, ecological inflation, and the rural/urban confounding that the low-exposure cohorts expose.
• ~20% that community water fluoridation at 0.7 mg/L causally reduces children's IQ by a detectable amount. Basis: the water dose-response being null below 1.5 mg/L, and null or positive results in the best-identified low-exposure cohorts.
• ~85% that, if any true effect exists at 0.7 mg/L, it is smaller than about one IQ point. Basis: the small point estimates and wide intervals in the individual-level data, and the null population cohorts.
• ~75% that fluoridation at recommended levels still reduces dental caries in contemporary populations, by a modest amount. Basis: the direction of the 2024 Cochrane estimates and three cessation studies; against it, low-certainty evidence and intervals that include no benefit.
• ~55% that the net dental benefit outweighs any plausible cognitive harm at recommended levels in a general population. Basis: a modest, uncertain dental effect against a small, unestablished cognitive effect; the comparison turns on baseline caries risk, access to dental care, and how much weight one gives to precaution.

Reversal conditions

This brief should be revised or withdrawn if any of the following occurs:

1. A large, prospective, low-risk-of-bias cohort with individual exposure biomarkers in an optimally fluoridated (≈0.7 mg/L) population reports a consistent inverse association with IQ that survives adjustment for socioeconomic position, other water contaminants, and nutrition.
2. Contamination of the low-exposure cohorts is demonstrated — for example, if the null U.S. and Swedish results are traced to exposure misclassification rather than the absence of an effect.
3. A plausible and replicated mechanism is established in humans at water concentrations at or below 1.5 mg/L, changing the prior on biological plausibility.
4. A meta-analysis restricted to low-risk-of-bias, individual-level, policy-relevant studies reports a pooled adverse effect materially larger than one IQ point.
5. Further contemporary evidence shows no caries benefit at all, which would leave the policy without a measurable benefit to weigh against even a small cognitive risk.
6. A randomized or sharp regression-discontinuity design assigns or sharply varies fluoridation and reports a cognitive endpoint.

References

1. National Toxicology Program. NTP Monograph on the State of the Science Concerning Fluoride Exposure and Neurodevelopment and Cognition: A Systematic Review. Research Triangle Park, NC: NTP; 2024. doi:10.22427/NTP-MGRAPH-8.
2. Taylor KW, Eftim SE, Sibrizzi CA, Blain RB, Magnuson K, Hartman PA, et al. Fluoride exposure and children's IQ scores: a systematic review and meta-analysis. JAMA Pediatrics. 2025;179(3):282–292. doi:10.1001/jamapediatrics.2024.5542.
3. Taylor KW, Eftim SE, Sibrizzi CA, et al. Addressing critiques of the evidence linking fluoride and children's IQ. Annals of Global Health. 2025;91(1):83. doi:10.5334/aogh.4853.
4. Kumar JV, Moss ME, Liu H, Fisher-Owens SA. Association between low fluoride exposure and children's intelligence: a meta-analysis relevant to community water fluoridation. Public Health. 2023;219:73–84. doi:10.1016/j.puhe.2023.03.011.
5. Veneri F, Vinceti M, Generali L, Giannone ME, Mazzoleni E, Birnbaum LS, et al. Fluoride exposure and cognitive neurodevelopment: systematic review and dose-response meta-analysis. Environmental Research. 2023;224:115239. doi:10.1016/j.envres.2023.115239.
6. Bashash M, Thomas D, Hu H, Martinez-Mier EA, Sanchez BN, Basu N, et al. Prenatal fluoride exposure and cognitive outcomes in children at 4 and 6–12 years of age in Mexico. Environmental Health Perspectives. 2017;125(9):097017. doi:10.1289/EHP655.
7. Green R, Lanphear B, Hornung R, Flora D, Martinez-Mier EA, Neufeld R, et al. Association between maternal fluoride exposure during pregnancy and IQ scores in offspring in Canada. JAMA Pediatrics. 2019;173(10):940–948. doi:10.1001/jamapediatrics.2019.1729.
8. Till C, Green R, Flora D, Hornung R, Martinez-Mier EA, Blazer M, et al. Fluoride exposure from infant formula and child IQ in a Canadian birth cohort. Environment International. 2020;134:105315. doi:10.1016/j.envint.2019.105315.
9. Malin AJ, Eckel SP, Hu H, Martinez-Mier EA, Hernandez-Castro I, Yang T, et al. Maternal urinary fluoride and child neurobehavior at age 36 months. JAMA Network Open. 2024;7(5):e2411987. doi:10.1001/jamanetworkopen.2024.11987.
10. Grandjean P, Meddis A, Nielsen F, Beck IH, Bilenberg N, Goodman CV, et al. Dose dependence of prenatal fluoride exposure associations with cognitive performance at school age in three prospective studies. European Journal of Public Health. 2024;34(1):143–149. doi:10.1093/eurpub/ckad170.
11. Broadbent JM, Thomson WM, Ramrakha S, Moffitt TE, Zeng J, Foster Page LA, et al. Community water fluoridation and intelligence: prospective study in New Zealand. American Journal of Public Health. 2015;105(1):72–76. doi:10.2105/AJPH.2013.301857.
12. Aggeborn L, Öhman M. The effects of fluoride in drinking water. Journal of Political Economy. 2021;129(2):465–491. doi:10.1086/711915.
13. Warren JR, Rumore G, Kim S, Grodsky E, Muller C, Manly JJ, Brickman AM. Childhood fluoride exposure and cognition across the life course. Science Advances. 2025;11(47):eadz0757. doi:10.1126/sciadv.adz0757.
14. Warren JR, Rumore G, Sicinski K, Herd P, Engelman M. Municipal water fluoridation, adolescent IQ, and cognition across the life course: evidence from the Wisconsin Longitudinal Study. Proceedings of the National Academy of Sciences USA. 2026;123(16):e2536005123. doi:10.1073/pnas.2536005123.
15. Marques RP, Bastos JL, Magalhães R. Community water fluoridation and intelligence quotient: systematic review and meta-analysis of observational studies. Ciência & Saúde Coletiva. 2026;31(2):e19682023. doi:10.1590/1413-81232026312.19682023.
16. Iheozor-Ejiofor Z, Walsh T, Lewis SR, Riley P, Boyers D, Clarkson JE, et al. Water fluoridation for the prevention of dental caries. Cochrane Database of Systematic Reviews. 2024;10(10):CD010856. doi:10.1002/14651858.CD010856.pub3.
17. McLaren L, Patterson S, Faris P, Chen G, Faris B, Rawal S, et al. Fluoridation cessation and children's dental caries: a 7-year follow-up evaluation of Grade 2 schoolchildren in Calgary and Edmonton, Canada. Community Dentistry and Oral Epidemiology. 2022;50(5):391–403. doi:10.1111/cdoe.12685.
18. Meyer J, Margaritis V, Mendelsohn A. Consequences of community water fluoridation cessation for Medicaid-eligible children and adolescents in Juneau, Alaska. BMC Oral Health. 2018;18:211. doi:10.1186/s12903-018-0684-2.
19. Yazdanbakhsh E, Bohlouli B, Patterson S, Amin M. Community water fluoride cessation and rate of caries-related pediatric dental treatments under general anesthesia in Alberta, Canada. Canadian Journal of Public Health. 2024;115(2):305–314. doi:10.17269/s41997-024-00858-w.
20. Kumar JV, Levy SM, Warren JJ. An update on community water fluoridation, part 2: fluoride exposure and children's IQ scores — relevance of systematic reviews and meta-analyses to community water fluoridation. Journal of the American Dental Association. 2026;157(5):471–486. doi:10.1016/j.adaj.2026.02.010.
21. Congressional Research Service. The Development of Federal Recommendations and Regulations for Fluoride in Drinking Water. CRS Report R48539; May 16, 2025. https://www.congress.gov/crs_external_products/R/PDF/R48539/R48539.2.pdf.
22. Iheozor-Ejiofor Z, Worthington HV, Walsh T, O'Malley L, Clarkson JE, Macey R, et al. Water fluoridation for the prevention of dental caries. Cochrane Database of Systematic Reviews. 2015;(6):CD010856. doi:10.1002/14651858.CD010856.pub2.