Ultra Processed Foods: the Government Has No Definition, and Said So

In July 2025 the Food and Drug Administration published a sentence that almost no article about ultra-processed food will quote: “Currently, there is no single authoritative definition for ultra-processed foods for the U.S. food supply.” The agency was not being coy. It said this while announcing, jointly with the USDA, a public request for information whose purpose was to help build one.

So the term at the centre of the loudest nutrition argument of the decade has no official American meaning. That does not make it useless, and it does not make the concern imaginary. It does mean that almost every confident claim you have read about ultra-processed food rests on a definition somebody chose. Then, in January 2026, the government did something stranger: it told Americans to avoid this food anyway, and said in the same breath that it was not defining it. Below: what changed in January 2026 and what did not, where the 55% figure comes from and what it is actually measuring, why adult consumption is falling rather than rising, what the government’s own advisory committee concluded and how weak it said the evidence was, which viral claims have no federal source at all, and the numbers that do exist and are worth acting on.

Where the definition actually comes from

The classification everyone is using is called Nova, developed by researchers in Brazil. It sorts food into four groups by degree of processing, with group four — ultra-processed — covering industrial formulations built from ingredients and additives not typically found in a home kitchen. MedlinePlus, the National Library of Medicine’s consumer health encyclopedia, explains the four groups to readers in roughly those terms.

Here is the part worth slowing down for. Nova is on US federal websites, and US federal researchers use it — the CDC computes the national statistic with it, and the USDA’s research arm built a study menu with it. But no US agency has adopted it as a definition of anything. The FDA’s own page on ultra-processed foods does not mention Nova at all. And the term “ultra-processed” appears in the Dietary Guidelines for Americans in neither the 2020–2025 edition nor the 2025–2030 edition that replaced it in January 2026.

Nova is therefore the working tool of American nutrition research and simultaneously not an American definition. Both things are true, and keeping them in view is what separates an honest account of this subject from an alarming one.

January 2026: the advice changed, the definition did not

While this argument was running, the document at the centre of it was replaced. The current edition of the Dietary Guidelines for Americans is the 2025–2030 edition, published in January 2026 by the Department of Health and Human Services and the Department of Agriculture, and one of its sections is headed “Limit Highly Processed Foods, Added Sugars, & Refined Carbohydrates”.

That heading is new, and it is worth being exact about what is new in it. The section tells readers to “avoid highly processed packaged, prepared, ready-to-eat, or other foods that are salty or sweet, such as chips, cookies, and candy that have added sugars and sodium (salt)”. Under sodium it says “Highly processed foods that are high in sodium should be avoided”. The government’s own site for the new edition puts the change plainly: “For the first time, official U.S. guidance calls on Americans to avoid highly processed food.”

The very next sentence on that page says what the instruction is not: “This is not an attack on industry or a legal definition.” The site’s FAQ repeats it — “We are not making a legal definition or attacking an industry”.

So the position, as of this edition, is this. American dietary guidance now tells you to avoid a category of food. It calls that category “highly processed”, not “ultra-processed”; the string “ultra” does not appear in the document at all. It gives no definition of the category, no criteria, no list of markers, and it does not cite Nova. And the departments that publish it have said in writing that they are not making a legal definition.

None of that contradicts what the FDA said in July 2025. It is the same fact with an instruction attached: the United States is now advising against something it has declined to define. That is a real change in advice and no change at all in the evidence — which is why the rest of this article still holds.

How much of the American diet — and the direction nobody reports

The number exists and it is recent. The CDC’s National Center for Health Statistics published it in Data Brief No. 536 in August 2025: during August 2021 to August 2023, the mean percentage of total calories from ultra-processed foods among Americans aged 1 and over was 55.0%. Young people aged 1–18 were higher at 61.9%; adults 19 and over lower at 53.0%. Among adults, consumption was lowest in the highest family income group.

Four of the top five sources of those calories were sandwiches including burgers, sweet bakery products, savoury snacks and sweetened drinks — which is to say the categories most people would have guessed.

Now the finding that contradicts the entire genre. Adult consumption has gone down. The same brief reports that the mean percentage of calories from ultra-processed foods among adults decreased from 55.8% in 2013–2014 to 53.0% in August 2021–August 2023. Nearly three percentage points over roughly a decade, in the wrong direction for a story about a problem spiralling out of control.

Two honest caveats belong with that 55%. It is a Nova number, not an FDA number. And the FDA’s own figures, quoted on its July 2025 announcement, are different ones: it estimates that approximately 70% of packaged products in the US food supply are “foods often considered ultra-processed”, and that children get over 60% of their calories from such foods. Note the construction — “often considered”, not “are”. That phrasing is the missing definition showing through in the agency’s prose.

What the advisory committee actually concluded

The most-cited American finding on ultra-processed food and body weight comes from the Scientific Report of the 2025 Dietary Guidelines Advisory Committee. The distinction matters: the advisory committee is a body that reviews evidence and reports to the departments. Its report is not the Dietary Guidelines, and the Dietary Guidelines themselves — both the edition current when the report was written and the one that replaced it — do not use the term.

What the committee concluded, for both children and adults, is that dietary patterns with higher amounts of food classified as ultra-processed are “associated with greater adiposity”. Three things in that sentence are load-bearing. Associated with is observational language, not causation. Greater adiposity is not the same as obesity. And classified as points at the problem underneath.

Because the committee attached a grade to that conclusion, and the grade is “Limited” — its own weakest tier above unassignable. That is the state of the American evidence base on the claim you have seen stated as settled fact a hundred times.

The committee also explained why. The studies it reviewed did not agree on what they were measuring: several methods exist for characterising ultra-processed foods, so the foods and dietary patterns across the reviewed articles were not consistent, and the definition in each case was the one its own authors chose. It noted, too, that standard dietary questionnaires do not reliably distinguish a home-cooked meal from a frozen dinner.

Read that beside the FDA sentence at the top of this article and you have two separate federal processes arriving at the same place from different directions. The agency says there is no authoritative definition; the committee says the research it reviewed did not have one either.

One more thing the committee said, and it is routinely misreported. It recommended that future committees examine ultra-processed foods in relation to type 2 diabetes, cardiovascular disease, cancer and cognitive decline. That is a list of things not yet examined in this process — the opposite of a set of findings about them.

The claims that have no federal source

That ultra-processed food is addictive. No US federal health authority classifies it that way. The nearest federal material describes people eating more calories, which is a different claim.

That it causes cancer or dementia. There is no US federal conclusion on either. The advisory committee listed both among outcomes it thinks should be looked at in future. The FDA’s own formulation is worth copying exactly: studies “have found links between the consumption of foods often considered ultra-processed with numerous adverse health outcomes”, including cardiovascular disease, type 2 diabetes, cancer, obesity and neurological disorders. Links, not causes — and “often considered”, not “are”.

That emulsifiers destroy the gut microbiome. We could not find this on any federal page. The only federal statement about emulsifiers we located is a functional one, from MedlinePlus: they prevent liquid products from separating.

That America permits what Europe bans. No federal source makes or rebuts that comparison. What is on the record cuts both ways: synthetic color additives must undergo batch certification, in which the FDA analyses a sample of each batch before it can be used — and the US position is changing anyway, since the agency moved against several dyes during 2025.

That last point deserves its own paragraph, because it is the clearest example on this subject of a headline losing half its source. In January 2025 the FDA revoked the authorisation for Red No. 3. The basis was the Delaney Clause, enacted in 1960, which prohibits authorisation of a food or colour additive if it has been found to induce cancer in humans or animals. On the same page announcing it, the agency also stated that the way Red No. 3 causes cancer in male rats does not occur in humans. Both halves are the FDA’s. Only the first half traveled.

An absence of official statements is not proof that something is harmless, and none of this is an argument that packaged food is fine, or a substitute for the basics of a decent diet. It is an argument for knowing which of the things you have been told has a government behind it.

The numbers that do exist

Here is the practical turn. “Avoid ultra-processed food” cannot be measured, because the category cannot be defined. But the Dietary Guidelines do carry limits for the things highly processed food tends to be high in — and in the 2025–2030 edition those limits became more specific, not less:

Note what those limits name: sugar, fat, sodium. Nutrients, not processing. And note how they changed. The 2020–2025 edition capped both added sugars and saturated fat the same way, at “less than 10 percent of calories per day starting at age 2”. The current edition keeps a 10% ceiling for saturated fat, drops the age qualifier, and replaces the added-sugar percentage with something you can check against a single meal. The limits got more concrete while the category they are aimed at stayed undefined.

And this is where the advisory committee’s own reasoning closes the loop. Explaining why the ultra-processed patterns in its review looked worse, it observed that many of the foods designated ultra-processed were higher in saturated fat, sodium and added sugars than the foods in the comparison patterns. The mechanism it points to is the three things that already have numbers.

Which means the actionable advice exists. It simply is not phrased as “avoid ultra-processed food” — it is phrased as three limits you can check on a label. Our guides to daily sodium and to added sugar and carbohydrates go through each in detail.

Two federal studies that do not agree, and why that is the interesting part

In 2023 the USDA’s Agricultural Research Service reported something that should be better known. Researchers at its Grand Forks centre built a seven-day, 2,000-calorie menu meeting the Dietary Guidelines recommendations then in force in which, using the Nova scale, 91% of calories came from ultra-processed foods. It scored 86 out of 100 on the Healthy Eating Index-2015. The building blocks were things like canned beans, instant oatmeal, ultra-filtered milk, whole wheat bread and dried fruit. The researchers were candid about the limits: the menu exceeded sodium recommendations and fell short on whole grains, and they noted that observational research associates ultra-processed products with adverse outcomes and that intervention studies are still needed.

Against that, a National Institutes of Health controlled feeding trial published in 2019. Twenty adults lived at the NIH Clinical Center for a month, spending two weeks on an ultra-processed diet and two on an unprocessed one, with both matched for calories, sugar, fat, fiber and macronutrients on offer and free access to eat as much as they liked. On the ultra-processed diet they ate about 500 calories more per day and gained around two pounds, losing the equivalent on the unprocessed one. The lead investigator described it as the first demonstration of causality. The release also says the study was small and that slight differences in protein between the two diets could potentially explain as much as half the difference in calorie intake.

Neither study refutes the other, and both are federal. One shows a diet built almost entirely from ultra-processed foods can meet the guidelines on paper. The other shows that when people are handed ultra-processed food and left to eat freely, they eat substantially more of it. The reconciliation is not a contradiction at all: the problem may be less what ultra-processed food is made of than how much of it people eat when it is in front of them. That is a more useful idea than either headline on its own.

Nobody else has a definition either

It is tempting to read the American gap as an American failure. It is not. The World Health Organization maintains a page for a Guideline Development Group on ultra-processed foods — a group whose composition is still described as provisional, pending confirmation by its guidelines review committee. As of writing, the WHO is assembling the panel that would produce a position, not publishing one.

The Pan American Health Organization, a WHO regional body, is further along and blunter, arguing that marketing of ultra-processed and processed products encourages excessive consumption. Worth noticing what its own sentence blames: energy, sugars, fats, saturated fats, trans fats and sodium. Even the most forward-leaning official body on this subject routes the harm through nutrients — the same ones with numbers attached.

Check a product yourself

NOVA is a research classification, not a federal definition, but it is the one most studies on processed food use. Search any packaged food by name, brand or barcode to see its NOVA group, the processing markers in its ingredient list and its additives.

This content is for educational purposes only and is not medical advice. Always consult a qualified healthcare professional for personalised guidance.