The Behavioral Connection
The most studied concern with food dyes is their effect on children's behavior, particularly hyperactivity and attention problems. The landmark study here is the 2007 Southampton trial [1], a randomized, double-blind, placebo-controlled study funded by the UK Food Standards Agency. Researchers gave 3-year-olds and 8/9-year-olds drinks containing either a mix of artificial colors plus sodium benzoate, or a placebo. Children consuming the additive mixes showed significantly increased hyperactive behavior compared to placebo, as measured by both parental reports and objective testing.
This wasn't the first study to find this link. A 2011 meta-analysis in Pediatrics found that removal of artificial food colors from the diet produced roughly one-third to one-half the effect of methylphenidate (Ritalin) on ADHD symptoms [5]. That's a meaningful effect size, especially for a dietary change with zero downside.
However, the evidence has important limitations. Not all children appear equally sensitive - some show strong reactions while others show none. The Southampton study used a mixture of dyes and sodium benzoate, making it difficult to isolate which specific additive drove the effect. And the behavioral changes, while statistically significant, were modest in absolute terms for the average child.
The FDA convened an advisory committee in 2011 to review this evidence and voted 8-6 that a causal relationship between dyes and hyperactivity had not been established in the general population, though they acknowledged effects in susceptible subpopulations. The EFSA took a more precautionary stance, leading to the EU warning labels [3].
Cancer and Other Concerns
Beyond behavior, some food dyes have raised concerns about carcinogenicity, though the evidence is less consistent [2] [4].
Red 3 (Erythrosine) was shown to cause thyroid tumors in rats at high doses, and the FDA banned it from cosmetics and externally applied drugs in 1990 - but never from food. This inconsistency persisted for over 30 years until California acted.
Red 40, the most widely consumed dye in the US, has shown some evidence of causing DNA damage in colon cells in animal studies, but results across studies are inconsistent. The EFSA re-evaluated it in 2009 and maintained its approval with an acceptable daily intake, while noting some data gaps [3].
Yellow 5 has shown genotoxicity in some in vitro studies, though animal feeding studies have generally not confirmed carcinogenicity [2]. It is also a known trigger for allergic-type reactions in a small percentage of people, particularly those with aspirin sensitivity.
Yellow 6 has shown evidence of adrenal tumors in animal studies, though again the evidence is disputed and the FDA has not found it sufficient to warrant regulatory action.
The honest assessment: the carcinogenicity data for most approved food dyes is a patchwork of conflicting studies, many using doses far exceeding normal human consumption. The behavioral evidence in children is stronger and more directly relevant to real-world exposures. The fundamental question is whether purely cosmetic ingredients deserve the benefit of the doubt when there are safer alternatives readily available - the EU has essentially answered no, while the US FDA continues to answer yes.
Regulatory Divergence and Study Quality
The disagreement between the FDA and EFSA on food dyes reflects broader philosophical differences in food safety regulation. The FDA generally requires strong proof of harm before restricting an approved additive, while EFSA leans more toward precaution, particularly when the additive serves no functional purpose beyond aesthetics.
The Southampton study [1] remains the highest-quality trial in this space. Its strengths: large sample (n=153 for 3-year-olds, n=144 for 8/9-year-olds), randomized double-blind placebo-controlled design, community-based population (not just children with diagnosed ADHD). Its weaknesses: the mixtures tested combined multiple dyes with sodium benzoate, effect sizes varied by age group and assessment method, and the 8/9-year-old results were less consistent than the younger group's.
Subsequent meta-analyses have generally confirmed a small but real effect of artificial food colors on behavior, with estimated effect sizes (Cohen's d) ranging from 0.21 to 0.44 depending on the analysis and whether the assessment was by parents (larger effect) or blinded observers (smaller effect) [5]. These are considered small to medium effects in behavioral research.
On carcinogenicity, much of the evidence base relies on studies conducted in the 1970s-1990s with methodological standards that would not meet current guidelines [4]. The EFSA has called for updated studies for several dyes but has not received them from manufacturers. This data gap itself is worth noting: the safety case for many dyes rests partly on the absence of evidence rather than evidence of absence.
One area of emerging research is the effect of food dyes on the gut microbiome. Preliminary studies suggest that some azo dyes (which include Red 40 and Yellow 6) are metabolized by gut bacteria into aromatic amines, some of which are known mutagens. This pathway needs more research but could explain some of the inconsistencies in whole-animal studies.