Peak Human
    Public Health · Evidence Review

    Is Fluoride Safe? Two Questions, Not One

    Toothpaste and drinking water are different exposures — and the evidence, the risk and the ethics differ with them

    Reviewed by Dr. Sanjeev GoelSeptember 21, 202612 min read

    Almost every public argument about fluoride collapses two separate questions into one. A tube of toothpaste and a municipal water supply are not the same exposure, and the evidence behind them is not the same evidence.

    The 60-Second Answer

    Fluoride toothpaste has a strong evidence base and a favourable risk–benefit ratio at any plausible exposure: 1,000–1,500 ppm reduces decay, and going higher adds essentially nothing. Community water fluoridation is a different question. Its modern benefit is real but much smaller than it was before fluoride toothpaste became universal — contemporary studies show a difference of roughly a quarter of a tooth, with a confidence interval that includes no benefit — and its clearest value today is in narrowing inequity, which is what Calgary's fourteen years without it suggest. On safety, the neurodevelopmental signal that reopened this debate sits above 1.5 mg/L, well above the 0.7 mg/L used in Canada and the United States. The best-quantified harm at 0.7 ppm is dental fluorosis — mostly faint white flecking — affecting roughly 12% of people at a level of aesthetic concern.

    Three serious documents landed close together

    Fluoride became contested again not because of a scandal, but because three careful documents arrived within months of each other: the US National Toxicology Program's 2024 monograph on fluoride and neurodevelopment[3], the 2024 Cochrane update on water fluoridation[1], and a 2025 JAMA Pediatrics meta-analysis of fluoride and children's IQ[4]. Each drew substantive published criticism.[7][14] Policy then moved faster than the evidence did.

    The useful response is neither “safe and effective, end of discussion” nor “neurotoxin in the water supply.” The evidence is more specific than either slogan, and its meaning changes with the exposure being discussed.

    Two different questions, not one

    This is the single most useful distinction in the topic. Toothpaste is topical, opt-in and concentrated. It is applied to the tooth surface and spat out; the dose that reaches the body is small and controllable. Water is systemic, population-wide and low-concentration. It is swallowed and it is not opt-in.

    Those are different evidence bases, different risk profiles and different ethical questions. Conflating them is the central error in most coverage in both directions: evidence for toothpaste does not settle water policy, and concern about swallowed fluoride does not erase toothpaste's topical benefit.

    What fluoride toothpaste actually does

    Cochrane's 2019 network meta-analysis included 96 randomised trials published from 1955 to 2014.[2] Against non-fluoride toothpaste in children and adolescents, 1,000–1,250 ppm gave an SMD of −0.28 (95% CI −0.32 to −0.25; 55 studies; high/moderate certainty), while 1,450–1,500 ppm gave an SMD of −0.36 (95% CI −0.43 to −0.29; 4 studies).

    Head to head, 1,450–1,500 ppm slightly beat 1,000–1,250 ppm (SMD −0.08, 95% CI −0.14 to −0.01; 10 studies). Above that, the benefit plateaued: 1,700–2,200 ppm versus 1,450–1,500 ppm was SMD 0.04 (95% CI −0.07 to 0.15), and 2,400–2,800 ppm was SMD −0.05 (95% CI −0.14 to 0.05).

    In young children's primary teeth, 1,500 ppm versus non-fluoride produced an MD of −1.86 dfs (95% CI −2.51 to −1.21). In adults, 1,000–1,100 ppm versus non-fluoride produced an MD of −0.53 DMFS (95% CI −1.02 to −0.04). The teaching point is simple: it works, and more is not better past about 1,500 ppm.

    Figure 1

    Fluoride toothpaste: the benefit plateaus above ~1,500 ppm

    Standardised mean difference in caries increment (negative favours the first-named group)

    Bars below zero favour the first-named concentration. Fluoride toothpaste clearly beats non-fluoride toothpaste, and 1,450–1,500 ppm slightly beats 1,000–1,250 ppm. Above roughly 1,500 ppm the confidence intervals include zero — the last two comparisons are statistically indistinguishable from no difference.

    Source: Walsh et al., Cochrane Database of Systematic Reviews 2019 (96 randomised trials); D(M)FS in children and adolescents.

    What water fluoridation does now

    Cochrane's 2024 update included 157 studies, all non-randomised, and stratified results around 1975 because that is when fluoride toothpaste became widespread.[1] Restricted to contemporary evidence, initiation of fluoridation gave a change in dmft of MD 0.24 (95% CI −0.03 to 0.52; 2 studies; 2,908 children; low certainty) in favour of fluoridation — about a quarter of a tooth, with an interval that includes no benefit. Caries-free proportions improved by about 4 percentage points in primary dentition and 3 points in permanent dentition, but both intervals crossed zero. Pre-1975 studies showed larger effects at very low certainty.

    Be straight about what this means: the modern marginal benefit is small and uncertain, largely because fluoride toothpaste now does most of the work. A 2026 British Dental Journal critique argues that Cochrane's evidence hierarchy, built around randomised trials, is the wrong instrument for a population-level public-health programme and excludes much relevant observational evidence.[7] Readers should know the review is contested by serious people, and why.

    The equity argument and the Calgary experiment

    The strongest modern case for fluoridation is also the most locally relevant. Calgary stopped fluoridating in 2011; Edmonton continued.[9] Seven years after cessation, social inequities in untreated decay had widened specifically in Calgary. The association between having no dental insurance and untreated decay was stronger in Calgary in 2018/19 than before cessation (OR for comparison of coefficients 1.89, 95% CI 1.36–2.63, p<0.001), and stronger than in still-fluoridated Edmonton (OR 1.44, 95% CI 1.03–2.02, p=0.033).[5]

    An earlier analysis of the same cohorts found inequities by insurance status and neighbourhood deprivation already more apparent in 2013/14 than before cessation.[6] Calgary resumed fluoridation on 30 June 2025 at 0.7 mg/L after a 2021 plebiscite in which 62% voted in favour.[9] Fluoridation's value today may be less about the population average than about children with no dental insurance and no ready access to a dentist.

    The neurodevelopmental question, handled carefully

    The NTP concluded with moderate confidence that higher fluoride exposures — explicitly framed as drinking-water concentrations exceeding the WHO guideline of 1.5 mg/L — are consistently associated with lower IQ in children; 18 of 19 high-quality studies showed an inverse association. It found low confidence for effects on adult cognition and said more studies are needed to understand whether lower exposures matter.[3][10]

    The 2025 JAMA Pediatrics meta-analysis pooled 74 studies — 45 from China and 52 rated high risk of bias — and found a pooled SMD of −0.45 (95% CI −0.57 to −0.33) across 59 studies.[4] The dose nuance matters. In the 31 studies measuring fluoride in drinking water, the inverse association held when exposed groups were restricted to below 4 mg/L and below 2 mg/L, but was null below 1.5 mg/L. In the subset of low-risk-of-bias studies, however, the association remained inverse even below 1.5 mg/L, and analyses using urinary fluoride remained inverse at every threshold.

    Individual-level data showed an IQ decrease of 1.63 points per 1 mg/L of urinary fluoride (95% CI −2.33 to −0.93), or 1.14 points (95% CI −1.68 to −0.61) in low-risk-of-bias studies. The authors themselves flagged limited data and real uncertainty in that low range, and the journal published correspondence challenging the data and analyses, to which the authors replied.[14]

    For a North American reader on 0.7 mg/L water, the clearest signal lives at concentrations roughly twice that or higher. The low-exposure range is genuinely uncertain rather than reassuringly clean, and most of the literature comes from high-fluoride endemic regions where total exposure bears little resemblance to a fluoridated Canadian city.

    Figure 2

    IQ points per 1 mg/L of urinary fluoride

    Individual-level analyses, with 95% confidence intervals

    Urinary fluoride is a measure of total intake from all sources, not of tap-water concentration. Across drinking-water studies the association was inverse below 4 mg/L and below 2 mg/L but null below 1.5 mg/L, although it persisted below 1.5 mg/L in the higher-quality subset — 45 of the 74 studies were conducted in China and 52 were rated high risk of bias.

    Source: Taylor et al., JAMA Pediatrics 2025;179(3):282–292.

    Dental fluorosis: the harm we can actually count

    From Cochrane 2024, at a fluoride level of 0.7 ppm, roughly 12% of people had fluorosis of aesthetic concern (95% CI 8–17%; 40 studies; 59,630 participants) and roughly 40% had fluorosis of any level (95% CI 35–44%; 90 studies; 180,530 participants).[1] Mostly this means faint white flecking rather than the brown mottling people picture.

    It is a cosmetic outcome, it is the best-quantified harm in the field, and it is honest to call it a real cost. Skeletal fluorosis is a genuine high-exposure disease and is very rare at Canadian concentrations.[8]

    Figure 3

    Dental fluorosis at 0.7 ppm

    Prevalence, % of participants, with 95% confidence intervals

    At the concentration used in Canada and the United States, roughly 40% of people show some fluorosis and about 12% show fluorosis at a level of aesthetic concern — in most cases faint white flecking rather than visible mottling. This is the best-quantified harm in the fluoridation literature.

    Source: Iheozor-Ejiofor et al., Cochrane Database of Systematic Reviews 2024 (90 studies, 180,530 participants for any fluorosis; 40 studies, 59,630 for aesthetic concern).

    Where the exposure actually comes from

    Water is one input, not the whole dose. In a fluoridated area the dominant controllable source in young children is swallowed toothpaste; tea, some foods, and infant formula reconstituted with fluoridated tap water also contribute. Health Canada sets the optimal concentration at 0.7 mg/L and a maximum acceptable concentration of 1.5 mg/L;[8] the WHO guideline value is also 1.5 mg/L.[10]

    Private wells are the real blind spot. Natural fluoride in groundwater can exceed 1.5 mg/L, and nobody adjusts it for you. The useful frame is total exposure rather than one isolated number — the same approach we take in our guide to mercury exposure from seafood.

    Where policy stands as of September 2026

    As of September 2026, Health Canada's optimal level remains 0.7 mg/L with a maximum acceptable concentration of 1.5 mg/L,[8] and Calgary's 2025 resumption ran against the North American trend.[9] In the United States, Utah and Florida enacted statewide bans on adding fluoride to public water in 2025. No further statewide ban had been enacted as of September 2026, while similar bills failed or stalled in roughly nineteen other states across the 2025–2026 sessions.

    The CDC still recommends community water fluoridation and still cites 0.7 mg/L.[13] The EPA announced an accelerated review of fluoride in drinking water in January 2026 and released a protocol for its fluoride human-health toxicity assessment in August 2026; the enforceable federal maximum contaminant level remains 4.0 mg/L.[11]

    On the litigation, in September 2024 a federal district court ordered the EPA to act on fluoridation risk under TSCA. On 21 May 2026, the Ninth Circuit vacated that order on procedural grounds — holding that the district court had improperly “commandeered” the case by reopening the record — and remanded for the merits to be decided on the first trial record, with EPA's standing objections addressed.[12] No court has ruled fluoridation unsafe. The question is unresolved rather than settled in either direction.

    What this means in practice

    • Use fluoride toothpaste. 1,000–1,500 ppm is the evidence-backed range; above roughly 1,500 ppm the benefit plateaus, and higher-strength pastes are a clinical decision, not a default.
    • For children, a rice-grain smear under age 3 and a pea-sized amount from 3 to 6, brushed under supervision, to limit how much is swallowed.
    • Spit, don't rinse. Rinsing washes away the fluoride that does the work.
    • On a private well, test the water. Natural fluoride can exceed 1.5 mg/L, and that is the range where the neurodevelopmental signal actually lives.
    • If you reconstitute infant formula, consider low-fluoride water for that specific use, especially in the first year.
    • Think in total intake, not in the water number alone — toothpaste, tea, food and water together.
    • Going fluoride-free as a reflex trades a well-quantified benefit for a poorly-quantified worry. If you are weighing it for your own family, that conversation belongs with your dentist or physician, not with a comment section.

    Frequently asked questions

    Is fluoride in tap water safe?

    At the 0.7 mg/L used in Canada and the United States, the established effect is dental fluorosis, mostly faint white flecking, affecting about 12% of people at a level of aesthetic concern. The neurodevelopmental findings that reopened this debate come overwhelmingly from areas with concentrations above the WHO guideline of 1.5 mg/L. The honest position is that risk at 0.7 mg/L looks small but the low-exposure evidence is thinner than public messaging has traditionally implied.

    Should I use fluoride-free toothpaste?

    For most people, no. Fluoride toothpaste at 1,000–1,500 ppm has the strongest and most consistent evidence of anything in preventive dentistry, and it is a topical exposure you spit out. Going fluoride-free gives up a well-measured benefit to avoid a risk that has not been demonstrated at toothpaste-level exposure.

    Does fluoride lower IQ?

    At high exposures, the association is consistent: the NTP concluded with moderate confidence that drinking-water concentrations above 1.5 mg/L are associated with lower IQ in children, and a 2025 meta-analysis found a pooled effect of −0.45 standard deviations. Below 1.5 mg/L the drinking-water analysis was null overall, though it remained inverse in the higher-quality subset, and the authors were explicit that data in that range are limited. Most of the underlying studies come from high-fluoride regions and were rated high risk of bias.

    Is fluoridated water safe for babies and infant formula?

    Infants fed formula reconstituted with fluoridated water get a relatively high dose for their size, and this is the main scenario where the fluorosis risk is worth actively managing. Using low-fluoride water for some or all formula preparation is a reasonable, low-cost precaution; breastfeeding delivers very little fluoride. Discuss it with your own clinician.

    What about well water?

    Test it. Natural groundwater fluoride varies enormously and can exceed the 1.5 mg/L maximum acceptable concentration, which is precisely the range where the neurodevelopmental evidence is strongest. A well owner is the one person in this debate with a genuinely actionable, and genuinely unknown, number.

    Why did Calgary bring fluoride back?

    Calgary stopped in 2011 and resumed on 30 June 2025 at 0.7 mg/L, after 62% of voters supported reinstatement in a 2021 plebiscite. Research comparing Calgary with still-fluoridated Edmonton found that inequities in untreated decay — particularly for children without dental insurance — widened in Calgary after cessation.

    References

    1. [1]Iheozor-Ejiofor Z, Walsh T, Lewis SR, et al. Water fluoridation for the prevention of dental caries. Cochrane Database Syst Rev. 2024;10(10):CD010856.
    2. [2]Walsh T, Worthington HV, Glenny AM, Marinho VC, Jeroncic A. Fluoride toothpastes of different concentrations for preventing dental caries. Cochrane Database Syst Rev. 2019;3(3):CD007868.
    3. [3]National Toxicology Program. NTP monograph on the state of the science concerning fluoride exposure and neurodevelopment and cognition: a systematic review. NTP Monogr. 2024;(8).
    4. [4]Taylor KW, Eftim SE, Sibrizzi CA, et al. Fluoride exposure and children's IQ scores: a systematic review and meta-analysis. JAMA Pediatr. 2025;179(3):282–292.
    5. [5]McLaren L, Patterson SK, Faris P, et al. Fluoridation cessation and oral health equity: a 7-year post-cessation study of Grade 2 schoolchildren in Alberta, Canada. Can J Public Health. 2022;113(6):955–968.
    6. [6]McLaren L, McNeil DA, Potestio M, et al. Equity in children's dental caries before and after cessation of community water fluoridation. Int J Equity Health. 2016;15:24.
    7. [7]Walmsley AD, Lowry R, Al Rasheed A, et al. Critique of the review of 'Water fluoridation for the prevention of dental caries' published by the Cochrane Collaboration in 2024. Br Dent J. 2026;241(1):26–30.
    8. [8]Health Canada. Guidelines for Canadian Drinking Water Quality: Guideline Technical Document — Fluoride.
    9. [9]City of Calgary. Fluoride in Calgary's water.
    10. [10]World Health Organization. Guidelines for drinking-water quality: fourth edition incorporating the first and second addenda. Geneva: WHO, 2022.
    11. [11]US Environmental Protection Agency. EPA announces next step in gold standard review of fluoride to inform protective recommendations. 2026.
    12. [12]Food & Water Watch, Inc., et al. v. United States Environmental Protection Agency, et al., No. 25-384 (9th Cir. May 21, 2026).
    13. [13]Centers for Disease Control and Prevention. Community water fluoridation: frequently asked questions.
    14. [14]Taylor KW, Eftim SE, Rooney AA. Concerns about data and analyses used in assessing fluoride exposure and children's IQ scores—reply. JAMA Pediatr. 2025.
    Disclaimer: This article is for educational purposes only and is not medical or dental advice. It does not establish a physician–patient relationship. Decisions about fluoride for you or your children — especially for infants and young children — should be made with your own dentist or physician.
    Back to Blog