In July 2023 two United Nations bodies published assessments of the same sweetener within weeks of each other, and the headlines read like contradictions. The World Health Organization’s cancer research agency classified aspartame as “possibly carcinogenic to humans”. A joint expert committee on food additives concluded the acceptable daily intake should stay exactly where it has been since 1981. Both were reported accurately and neither was reported completely, because the two bodies were not answering the same question. We set out each side’s claims, the sources behind them, our star ratings of those sources, and where the real disagreement sits. Truza does not score the debate, only the sources.
What both sides accept
The common ground is wide, which the coverage rarely conveys. Both sides accept that aspartame is among the most studied food additives in existence. Both accept its basic chemistry: it breaks down in the digestive tract into phenylalanine, aspartic acid and methanol, none of which is unique to aspartame, and a can of diet cola yields less methanol than a glass of tomato juice. Both accept that it does not raise blood glucose, which is the reason it exists. Both accept the phenylketonuria warning, the one population for whom the label’s advisory genuinely matters. And both accept that typical consumption sits far below the safety limits, roughly 40 milligrams per kilogram of body weight internationally and 50 in the United States. A 355 ml can of diet cola contains around 180 milligrams. For a 70 kilogram adult, the international limit works out to about 2,800 milligrams a day, which is roughly 15 cans.
The case for concern
Those raising it rest on the International Agency for Research on Cancer, which in July 2023 placed aspartame in Group 2B, “possibly carcinogenic to humans”. The classification rests on two kinds of evidence the agency called “limited”: human studies showing an association between aspartame consumption and cancer, and animal studies showing cancer at high doses. The strongest recent source on this side is not activist literature, it is a 2022 prospective cohort study in the NutriNet-Santé population, published in PLOS Medicine, following about 102,000 French adults and reporting a modestly higher overall cancer risk among higher consumers of aspartame, with liver cancer the standout association. Supporting it are the Ramazzini Institute’s multi-generational rat studies from 2005 and 2010, which reported lymphomas and leukemias at doses the authors called comparable to human exposure.
Our rating: the NutriNet cohort earns three stars. It is prospective, large and peer reviewed, but it is observational, its dietary intake is self-reported, and the reported effect sizes sit near the range where residual confounding and multiple comparisons do their work. The Ramazzini studies earn two stars. They are independent, which is worth something, but regulators including the United States FDA and the European Food Safety Authority reviewed them and found design and interpretation problems, including a respiratory infection confounding the animal results. The IARC monograph itself earns four stars: a structured expert process, published methods, a transparent vote. Its weakness is scope, which we come to.
The case for safe at current levels
The other side rests on the risk assessors. The joint FAO/WHO expert committee on food additives met the same month, reviewed the same evidence plus a larger toxicology base, and reaffirmed the 40 milligram per kilogram limit, concluding that the human evidence did not support a hazard at actual consumption levels and that the animal effects appeared only at doses far beyond anything a person ingests. The FDA responded to the IARC classification by saying its scientists had “no safety concern” at current levels, that the evidence does not support the Group 2B conclusion, and that a person weighing 70 kilograms would need to drink more than a dozen cans of diet soda a day to exceed the limit. The European Food Safety Authority’s full re-evaluation in 2013 reached the same place, and the authority is currently re-examining it again.
Our rating: the risk assessments earn four to five stars. They are systematic reviews of the full dosing literature, dose-response and all, conducted by standing regulatory bodies with published criteria. The honest caveat cuts both ways, and each side names it: regulatory dossiers include industry-submitted studies, because food additive regulation is built on industry submissions. That is standard practice in every jurisdiction, and both sides are quick to point out the funding of the other’s studies. Watch for the funding argument being used as a substitute for the design argument, which it usually is.
Where the real disagreement sits
Read the two documents closely and the contradiction mostly dissolves. IARC answers a hazard question: can this substance cause cancer under any circumstances? Group 2B means the evidence for that is limited, and it is the same group that contains aloe vera whole leaf extract and traditional pickled vegetables. The risk assessors answer an exposure question: does it cause harm at the amounts people actually eat and drink? Neither body disputed the other’s finding, and both said so in their own releases.
What remains after that is a values disagreement dressed as a data disagreement. Should a regulator act on limited evidence pointing one way, or wait for the stronger evidence that may never arrive? That is a judgement about uncertainty tolerance, and it is the reason the two camps talk past each other in press coverage. The data dispute underneath it is narrow: whether the observed human associations are causal and whether the animal dose-response transfers. On those two questions the strongest sources on each side genuinely disagree, and a reader deserves to know both, a task the coverage mostly declined, in the same way it has with the contested territory we map in our pieces on ultra-processed foods and moderate drinking.
Nothing here is dietary advice. If you know a stronger named source on either side, submit it, and we will rate it the same way we rated these.
