Is fluoride bad for you? What the evidence actually shows at each dose

At the level added to drinking water in the United States, which is 0.7 milligrams per litre, there is no good evidence that fluoride harms you, and there is not enough evidence to be sure either. Above 1.5 milligrams per litre, a US government review found with moderate confidence an association with lower IQ in children. Those two sentences are the whole argument, and almost nobody states both.

What does fluoride in water actually do?

Fluoride makes tooth enamel more resistant to the acid produced by bacteria, and it helps repair early damage before it becomes a cavity. That mechanism is not disputed by anyone.

What has changed is how much extra benefit comes from putting it in the water, now that almost everyone also brushes with fluoride toothpaste twice a day.

The Cochrane review of water fluoridation was updated in October 2024 by Zipporah Iheozor-Ejiofor and colleagues. It pulled together 157 studies, all of them non-randomised, because you cannot randomise a city's water supply.

Looking only at studies from after 1975, when fluoride toothpaste became widespread, the reviewers found that water fluoridation may slightly increase the proportion of children with no tooth decay: about 4 percentage points for baby teeth and 3 for adult teeth. They rated that evidence low certainty, and said the estimates include the possibility of no benefit at all.

That is a far smaller figure than the 35 and 26 per cent reductions quoted from older reviews, and it is the number that triggered the current argument.

So is the benefit now zero?

No, and the review does not say that. It says the effect in the modern era looks small and the certainty is low, which is a different claim.

It also matters that the 2024 review has itself been formally criticised. In 2026 a group including A. Damien Walmsley published a critique in the British Dental Journal arguing that applying trial-style evidence criteria to public health programmes excludes a great deal of relevant information, and that newer causal inference methods for observational data would give a fairer picture.

That is a real methodological disagreement among dental researchers, not a political one, and it is unresolved.

How much fluoride is in US tap water?

Since May 2015 the US Public Health Service has recommended a single level of 0.7 milligrams per litre for water systems that add fluoride. That replaced a 1962 recommendation which varied by air temperature and ranged from 0.7 to 1.2 milligrams per litre.

The stated reason for lowering it was to keep the protection against decay while limiting dental fluorosis, the faint white mottling of enamel that comes with higher intake. The notice also acknowledged that water is now only one of several sources of fluoride.

Two other numbers matter and get confused with the first. The Environmental Protection Agency's enforceable maximum is 4.0 milligrams per litre, with a non-enforceable secondary standard of 2.0. The World Health Organization's guideline value is 1.5.

So 0.7 is a target for systems that fluoridate, not a limit, and the limits sit well above it.

What did the US government review actually find about IQ?

The National Toxicology Program published its monograph on fluoride and neurodevelopment on 21 August 2024.

Its conclusion was that higher fluoride exposure, given as an example of drinking water above 1.5 milligrams per litre, is associated with lower IQ in children, held with moderate confidence. It also said there were insufficient data to determine whether 0.7 milligrams per litre affects children's IQ, and it found no evidence of adverse effects on adult cognition.

The companion meta-analysis by Kyla Taylor and colleagues appeared in JAMA Pediatrics in 2025. It is worth reading the detail, because the detail is where both sides stop quoting.

It included 74 studies. Forty-five were conducted in China, 12 in India, and none in the United States. Fifty-two were rated high risk of bias and 22 low risk of bias.

Across 59 studies with group-level exposure measures, covering 20,932 children, the pooled standardised mean difference was -0.45. In the 31 studies that measured fluoride in drinking water, the association stayed inverse when exposure was restricted to under 4 and under 2 milligrams per litre, and then went null below 1.5. When the same analysis was limited to the low risk of bias studies, it stayed inverse below 1.5.

For urinary fluoride, 13 studies with individual-level measures found IQ scores 1.63 points lower per additional milligram per litre, falling to 1.14 points among the low risk of bias studies.

The authors' own conclusion names the gap plainly: there were limited data and uncertainty about the dose-response relationship when exposure was estimated from drinking water alone at concentrations below 1.5 milligrams per litre.

Why does the 1.5 threshold keep coming up?

Because it is roughly twice the US target and it is where the evidence becomes consistent.

Most of the studies finding an effect come from places with naturally high fluoride in groundwater, at concentrations that would not be legal to add anywhere in the US or UK. Extrapolating from 3 milligrams per litre in a Chinese village to 0.7 in Cincinnati is the step that the NTP explicitly declined to take, and the step that campaigners on both sides take anyway.

The honest position is that the low-exposure evidence is thin rather than reassuring. Thin evidence is not proof of safety. It is also not proof of harm.

Which US states have banned fluoride in water?

Two, both in 2025. Utah legislated first and Florida followed. As of late September 2026 those remain the only statewide bans, although bills have been introduced in a number of other states.

Fluoride has also come under pressure from other directions, including federal moves against ingestible fluoride supplements prescribed for children. Local councils have always been able to stop fluoridating, and many have.

Does this mean I should avoid fluoride toothpaste?

The evidence does not support that, and the two questions are different.

Toothpaste is a topical application that you spit out. Water fluoridation is swallowed, and the IQ studies are about ingested exposure measured in water or urine. The benefit of fluoride toothpaste for preventing decay is one of the better-established findings in dentistry, and the Cochrane review's own explanation for the shrinking benefit of water fluoridation is that toothpaste is now doing the work.

Supervised care matters more for children. Swallowing toothpaste is the main avoidable source of excess intake at young ages, which is why a smear rather than a ribbon is advised for under-threes. If you want to reduce ingested fluoride, that is the lever with an actual evidence base behind it, unlike oil pulling.

What should I actually do?

If you are on a public supply in the US or UK, your water is at or below 0.7 to 1.0 milligrams per litre and you can find the figure on your supplier's annual water quality report.

If you use a private well, test it. Natural fluoride in groundwater can be several times the recommended level, and that is the situation the NTP monograph actually describes. The EPA's enforceable 4.0 milligram limit does not apply to private wells.

If you are pregnant or have young children and you want to be cautious, the measures that follow from the evidence are checking your water concentration, not adding fluoride supplements unless a dentist has advised them, and supervising toothbrushing so the paste is not swallowed.

What the evidence does not support is a reverse-osmosis filter bought on the strength of a headline about Chinese villages, any more than it supports telling people the question has been settled. We take the same approach to other everyday exposures, from red 40 to seed oils: the dose and the quality of the studies decide the answer.

This article is general information and not medical or dental advice. Speak to a qualified clinician or dentist about your own circumstances.