Modern fluoridation studies show a benefit of about a quarter of a tooth

The most recent Cochrane review of community water fluoridation, published in October 2024, found that in studies conducted after fluoride toothpaste became widespread, starting fluoridation may produce a difference of about a quarter of a tooth in children's decayed, missing or filled primary teeth — an effect estimate that includes the possibility of both benefit and no benefit, on low-certainty evidence [s1]. Separately, a 2025 meta-analysis in JAMA Pediatrics found an inverse association between fluoride exposure and children's IQ that persisted at every drinking-water threshold it tested in low risk-of-bias studies, but was null below 1.5 mg/L when all studies of drinking-water fluoride were pooled [s2]. Both of those sentences are accurate, and neither is the version most often repeated.

What the Cochrane review found

The review included 157 studies, every one of them non-randomised [s1]. Cochrane downgraded the certainty of all the evidence for the inherent risks of those designs, particularly around management of confounding and blinding of outcome assessors, and downgraded some further for imprecision, inconsistency or both [s1]. Crucially, it stratified results according to whether data were collected before or after fluoride toothpaste came into widespread use in 1975 — because evidence from older studies may not apply to contemporary societies where most people already brush with fluoride [s1].

For fluoridation initiation, drawing on 21 studies, the contemporary post-1975 results were these. Change in decayed, missing or filled primary teeth: mean difference 0.24 (95% confidence interval −0.03 to 0.52; P = 0.09; 2 studies, 2,908 children), low certainty — which the review translates as approximately one-quarter of a tooth in favour of fluoridation, with the estimate spanning both benefit and no benefit [s1]. Contemporary data on permanent teeth (4 studies, 2,856 children) and tooth surfaces (1 study, 343 children) left the review very uncertain [s1].

On the proportion of children with no caries at all, fluoridation may produce a slightly greater change over time in primary dentition (MD −0.04, 95% CI −0.09 to 0.01; P = 0.12; 2 studies, 2,908 children) and permanent dentition (MD −0.03, 95% CI −0.07 to 0.01; P = 0.14; 2 studies, 2,348 children) — differences of 4 and 3 percentage points respectively, favouring fluoridation, again on low-certainty evidence that includes the possibility of no benefit [s1].

Where older studies were the only evidence available, the review was unsure of the size of any effect on every outcome [s1]. On cessation, one post-1975 study of 2,994 children left the review unable to determine whether stopping fluoridation affected decay [s1]. No contemporary data on adverse effects were available at all [s1]. One study reported on socioeconomic disparities and found no evidence that deprivation influenced the relationship between water exposure and caries status [s1].

The one adverse outcome the review quantified is dental fluorosis, carried forward from the previous version without a new search. At a fluoride level of 0.7 parts per million, approximately 12% of participants had fluorosis of aesthetic concern (95% CI 8% to 17%; 40 studies, 59,630 participants) and approximately 40% had fluorosis of any level (95% CI 35% to 44%; 90 studies, 180,530 participants) [s1]. On other adverse effects — skeletal fluorosis, bone fractures and skeletal maturity, from 5 studies with incomplete participant numbers — the review was unsure [s1].

What the IQ meta-analysis found, in full

The JAMA Pediatrics meta-analysis included 74 studies: 64 cross-sectional and 10 cohort [s2]. Most were conducted in China (45); the rest came from India (12), Iran (4), Mexico (4), Canada (3), Pakistan (2), and one each from Denmark, New Zealand, Spain and Taiwan [s2]. Fifty-two studies were rated high risk of bias and 22 low [s2]. Sixty-four reported inverse associations between fluoride exposure and IQ [s2].

Pooling 59 studies with group-level exposure measures covering 20,932 children gave a standardised mean difference of −0.45 (95% CI −0.57 to −0.33; P < .001) [s2]. Among the 31 studies measuring fluoride in drinking water, a dose-response association was found between exposed and reference groups (SMD −0.15, 95% CI −0.20 to −0.11; P < .001), and the association remained inverse when exposed groups were restricted to below 4 mg/L and below 2 mg/L — but was null at less than 1.5 mg/L [s2]. In analyses restricted to the low risk-of-bias studies, the association remained inverse at below 4 mg/L, below 2 mg/L and below 1.5 mg/L [s2].

Twenty studies measuring urinary fluoride showed an inverse dose-response association (SMD −0.15, 95% CI −0.23 to −0.07; P < .001) that held at all three thresholds and among low risk-of-bias studies [s2]. Thirteen studies with individual-level measures found an IQ decrease of 1.63 points (95% CI −2.33 to −0.93; P < .001) per 1-mg/L increase in urinary fluoride, and 1.14 points (95% CI −1.68 to −0.61; P < .001) among low risk-of-bias studies [s2].

The authors' own stated conclusion identifies the pivotal uncertainty: there were limited data and uncertainty in the dose-response association when fluoride exposure was estimated by drinking water alone at concentrations less than 1.5 mg/L [s2].

Where the exposure levels sit

Those thresholds only mean something set against real concentrations. In the United States, the EPA enforces a primary drinking-water standard for fluoride of 4.0 mg/L, and a secondary standard of 2.0 mg/L; systems that exceed 2 mg/L without exceeding 4.0 mg/L must notify the people they serve within 12 months [s4]. The Cochrane review's fluorosis estimates are pegged to 0.7 ppm [s1], the level at which community water fluoridation is typically operated.

A 2025 rapid systematic review in Advances in Nutrition examined 58 studies of non-dental health outcomes from fluoride in drinking water — blood pressure, neurological outcomes, bone cancers, thyroid function, skeletal outcomes, low birth weight and preterm birth, and metabolic outcomes [s3]. It found some evidence of adverse neurological and thyroid effects in children, and of adverse blood pressure, thyroid and skeletal effects in adults [s3]. Its statement of where those effects occurred is the most directly relevant sentence in the current literature: in general, at drinking water fluoride concentrations above 1.5 mg/L, with no evidence of adverse effects at concentrations below 1.0 mg/L in either children or adults [s3]. The review's overall conclusion was that no convincing evidence of harm was identified from community water fluoridation at concentrations between 0.7 and 1.0 mg/L, while uncertainty remains at higher concentrations [s3]. It also reported that the study designs and methodological limitations precluded causal inference, and that the most common risk of bias was insufficient control of confounders [s3].

What is and is not established

Three things are reasonably firm. The measured benefit of initiating fluoridation in a population that already uses fluoride toothpaste is small and uncertain, not the large effect the older literature recorded [s1]. Dental fluorosis is common at operating levels, mostly at levels below aesthetic concern [s1]. And the harm signals in the non-dental literature concentrate above 1.5 mg/L, more than twice the level at which fluoridation is run [s2] [s3].

What is not established is causation in either direction of the argument. Every study in the Cochrane review was non-randomised [s1]; the IQ literature is dominated by cross-sectional studies rated high risk of bias [s2]; and the non-dental review states outright that causal inference was precluded [s3]. The disagreements about fluoridation policy are real, but they are not disagreements that the current evidence base can settle by itself.

This article summarises published systematic reviews and regulatory limits. It is not medical or policy advice, and it does not describe conditions in any particular water system.