The FDA banned Red Dye No. 3 from cosmetics in 1990 because it caused thyroid tumors in male rats at high doses. Then, for 35 years, the agency allowed Americans to eat it. That asymmetry is the story. Not the chemistry, not the cancer statistics, not the consumer petitions. The fact that the same substance was deemed too risky to put on your skin but acceptable to swallow is a regulatory contradiction so stark it almost reads as satire.
The ban announced in 2025, effective 2027, finally closes that gap. Critics argue the human cancer evidence is weak, and they are right. No study has quantified U.S. cancer cases directly attributable to erythrosine. The animal data comes from male rats fed doses far exceeding typical human exposure, and the mechanism involves a rat-specific hormonal pathway that does not map cleanly onto human thyroid biology. These are legitimate scientific objections. I take them seriously.
When the Absence of Evidence Is Not Evidence of Safety
Here is the methodological problem with waiting for human proof: you cannot run a controlled trial where you feed people a suspected carcinogen for decades and measure outcomes. Epidemiological studies of food additives are notoriously difficult to isolate because people eat hundreds of compounds simultaneously, and confounders multiply faster than you can control for them. Think of it like trying to identify which instrument in a 60-piece orchestra is slightly out of tune while the whole ensemble is playing. The signal exists. Finding it cleanly is nearly impossible.
So the question is not whether we have confirmed human harm. The question is what standard of evidence we should require before removing a synthetic dye that serves no nutritional purpose whatsoever. Red No. 3 makes maraschino cherries redder. It makes certain candies more vivid. Beet extract and carmine can do the same job. The cost of removing it is cosmetic, literally. The cost of keeping it, if the animal signal eventually proves meaningful in humans, is not.
Dr. Michael Greger framed the absurdity well in 2026: the dye was banned from anything going on the skin over 30 years ago due to cancer risk, yet remained acceptable to eat. That is not a precautionary principle run amok. That is a regulatory agency finally applying consistent logic to its own prior decision.
The 35-Year Gap Is the Real Scandal
New York's 2026 Healthy Food Act now requires the food industry to demonstrate additive safety to regulators, rather than waiting for regulators to prove harm. That inversion matters enormously. The old model placed the evidentiary burden on public health agencies with limited resources, while manufacturers faced no deadline to generate safety data. The new model asks a simpler question: if you cannot show it is safe, why is it in the food supply?
I want to be precise about what I am not arguing. I am not claiming Red No. 3 causes cancer in humans. The evidence does not support that claim. What I am arguing is that the methodology for keeping a substance in food should require more than the absence of a smoking gun, especially when the same substance was removed from cosmetics on precautionary grounds a generation ago.
The FDA should hold the 2027 effective date. State-level bans in California and New York should proceed. And the food industry should treat New York's burden-shifting law as the template, not the exception. A dye that colors candy and has been banned from lipstick since 1990 does not deserve 35 more years of benefit of the doubt.
