Food dyes, hyperactivity and the Red No. 3 ban: what the evidence shows

Two different questions get tangled together whenever synthetic food dyes are in the news, and the evidence points opposite ways on each. On cancer, the US Food and Drug Administration's January 2025 decision to ban FD&C Red No. 3 rests on a finding in rats through a mechanism the agency itself calls rat-specific, not on evidence that the dye causes cancer in people [s1]. On children's behaviour — a claim often dismissed as folklore — the controlled-trial evidence is actually stronger, and it is what regulators have been slower to act on [s2][s3].

The Red No. 3 ban, and what it does and doesn't say

In January 2025 the FDA granted a petition and repealed the regulations that had permitted FD&C Red No. 3 (erythrosine) in foods, including dietary supplements, and in ingested drugs [s1]. The trigger was legal, not new toxicology: the petitioners supplied data showing the additive induces cancer in male rats, and under the Delaney Clause — the provision of US law that bars any colour additive found to cause cancer in humans or animals — the FDA concluded it had no discretion to keep the listing [s1]. The agency's own finding is that Red No. 3 "can induce cancer in male rats, through a rat specific hormonal mechanism," and it revoked the uses "as a matter of law" [s1]. The ban takes effect on 15 January 2027 for food and 18 January 2028 for ingested drugs, with the window for objections having closed on 18 February 2025 [s1].

So the honest summary is narrow: a decades-old rat finding finally forced a legally required removal, not a demonstration that Red No. 3 in a sweet harms a child. That distinction is the kind of thing sales copy and scare copy both blur, and it is exactly the terrain where the cancer question we cover in what raises cancer risk in how food is cooked also lives.

The behaviour question has better evidence

The stronger science concerns attention and activity. The landmark study is the 2007 "Southampton" trial, a randomised, double-blind, placebo-controlled crossover in which 153 three-year-olds and 144 eight- and nine-year-olds drank a challenge containing sodium benzoate plus one of two mixes of artificial colours, or a placebo [s2]. The outcome was a composite "global hyperactivity aggregate" built from observed behaviour and parent and teacher ratings [s2]. Among the three-year-olds, mix A produced a significant adverse effect versus placebo (effect size 0.20, 95% confidence interval 0.01 to 0.39; p=0.044), rising to 0.32 (0.05 to 0.60; p=0.02) when the analysis was limited to children who drank most of the juice [s2]. Among the older children, restricted to those consuming at least 85 percent of drinks, both mixes had significant effects: 0.12 (0.02 to 0.23; p=0.023) for mix A and 0.17 (0.07 to 0.28; p=0.001) for mix B [s2]. The authors concluded that artificial colours, or a sodium benzoate preservative, or both, increased hyperactivity in children in the general population [s2].

The effects are small and the confidence intervals sit close to zero, which is why the finding is genuinely contested rather than settled — but it is a real, replicated signal from a rigorous design, not an anecdote.

What a systematic review made of it

A 2022 review conducted for the California Office of Environmental Health Hazard Assessment pulled the human and animal evidence together [s3]. It identified 27 clinical trials of children exposed to synthetic dyes, 25 of them challenge studies; 16 of those 25 (64 percent) found some evidence of a positive association with adverse behaviour, and in 13 (52 percent) it was statistically significant [s3]. The review's pointed conclusion was about regulation: the FDA's current acceptable daily intakes rest on older studies that were not designed to detect these behavioural effects, and for four dyes with adequate dose-response data the basis of those intake limits "may not be adequate to protect neurobehavior in susceptible children" [s3]. That is a provenance point worth keeping in view — the reassuring safety thresholds were set to answer a different question than the one parents are asking.

How to read this

The two dye stories should not be merged. The Red No. 3 removal is a cancer decision compelled by a rigid statute on a rat-specific mechanism, and it says little about everyday human risk [s1]. The behaviour evidence is the more solid of the two: controlled trials show a small but repeatable increase in hyperactivity in some children, enough that the intake limits protecting them look dated [s2][s3]. It is a close cousin of the sugar-and-hyperactivity question, where the popular belief runs ahead of the trials — except that here the trials lend the belief more support. Readers weighing additives generally may also find the added-sugar evidence and what "organic" does and doesn't buy you useful. This article is informational and is not dietary or medical advice.