Ultra-Processed Food, Explained: What NOVA Means and How to Read an Ingredient List

Picture two boxes of oatmeal on a shelf, both with “whole grain” on the front. Turn them over and they may not be the same kind of food at all. One can be just oats. The other can be oats plus a list of ingredients and additives chosen to make it taste, look and keep a certain way.

That gap between the front of the pack and the back of it is what this article is about. Nobody sat you down and taught you to read the back. The front is written to sell; the back is there because the law requires it. Learning to read it is a low-cost way to take back a default that was set for you.

The short version: “ultra-processed” usually means the NOVA classification. Researchers at the University of São Paulo developed it, and a 2019 report published by the UN Food and Agriculture Organization (FAO) sets it out. FAO notes the report’s views are the authors’ own and not an endorsement of NOVA. NOVA sorts food by how and why it was processed, not by its calories. Its practical rule is simple: if the ingredient list contains at least one thing you would never or rarely use in a kitchen (maltodextrin, high-fructose corn syrup, protein isolates, hydrogenated oil) or a class of additive used only in ultra-processed products (flavours, colours, emulsifiers, sweeteners and similar), the product counts as ultra-processed. A tightly controlled NIH trial found people ate about 500 more calories a day on an ultra-processed diet. A large body of observational research links higher intake to worse health outcomes, but most of that evidence is rated low quality. And as of 28 October 2026 the United States has no published federal definition; a proposed one has been under White House (OMB) review since August 2026.

This article is educational and is not medical or nutritional advice. If you have diabetes, a history of disordered eating, a food allergy, or any condition managed by a clinician, talk to them before you change how you eat.

What NOVA actually is: four groups, sorted by processing

Most nutrition labels tell you what is in a food: grams of sugar, fat, salt. NOVA asks a different question: what was done to the food, and for what purpose? The 2019 FAO report (Monteiro and colleagues) sorts every food into one of four groups:

1. Group 1: unprocessed or minimally processed foods. Edible parts of plants and animals, plus fungi and water, sometimes dried, ground, roasted, pasteurised, chilled, frozen or vacuum-packed. Think plain oats, fresh or frozen vegetables, eggs, milk, plain meat, dried beans.
2. Group 2: processed culinary ingredients. Oils, butter, lard, sugar and salt. These are things you cook with, not things you eat on their own.
3. Group 3: processed foods. Group 1 foods with group 2 ingredients added: vegetables in brine, fruit in syrup, tinned fish in oil, simple cheeses, and most freshly baked bread. They usually have two or three ingredients and you can still recognise the original food.
4. Group 4: ultra-processed foods. In the report’s words, “formulations of ingredients, mostly of exclusive industrial use.” Whole foods are broken down into fractions (sugars, starches, oils, proteins), some of those are chemically modified, and the result is assembled with little or no whole food left, then made appealing with flavours, colours and emulsifiers.

The line between groups 3 and 4 is where most real decisions happen. The report gives a clean example: an industrial bread made only from wheat flour, water, salt and yeast is a processed food (group 3). The same bread with emulsifiers or colours in the list is ultra-processed (group 4).

The one rule that does most of the work

Food makers do not have to put their processing methods on the label. The FAO report says so directly, and admits that this sometimes makes ultra-processed foods hard to identify with confidence. Its workaround is the ingredient list, which pre-packaged food must carry by law. The rule:

• Look near the start and middle of the list for food substances never or rarely used in kitchens. The report names fructose, high-fructose corn syrup, fruit juice concentrate, invert sugar, maltodextrin, dextrose, lactose, soluble or insoluble fibre, hydrogenated or interesterified oil, hydrolysed proteins, soya protein isolate, gluten, casein, whey protein and mechanically separated meat.
• Look near the end of the list for additive classes used only in ultra-processed food. These are flavours, flavour enhancers, colours, emulsifiers, emulsifying salts, artificial sweeteners, thickeners, and foaming, anti-foaming, bulking, carbonating, gelling and glazing agents.
• One hit is enough. A single item from either list classifies the product as ultra-processed.

Preservatives on their own do not make something ultra-processed under NOVA. The report notes they also appear in processed foods and culinary ingredients.

What I checked on 27 October 2026, and what it showed

Rather than describe the rule in the abstract, I applied it to real labels. On 27 October 2026 I pulled ingredient lists from USDA FoodData Central, the US Department of Agriculture’s public food database, where brand owners submit label data. I classified each one by the FAO rule above. Brand names are left out on purpose; the FDC ID lets you look up the exact record yourself.

Three things stood out:

• The two flavoured oatmeals land in the same group but are not equally processed. One has two markers: “natural flavor” and a monk fruit sweetener. The other has five. NOVA is binary at that line: two markers or five, both are group 4. That is a real limit of the system, covered below.
• “Whole grain” on the front tells you nothing about the group. Both flavoured oatmeals list whole grain oats first.
• The bread shows FAO’s own example in the wild. Flour, water, yeast and salt would be group 3. Corn syrup and three emulsifiers move it to group 4.

What this check does not show: the dates are when FDC published each record, the underlying label data may be older, and current packs may be reformulated. I did not test, taste, or eat these products, and nothing here says any one of them is harmful to you. The point is the method, not a verdict on a box.

What the evidence shows, and how strong it is

A small, tightly controlled trial

In 2019, Kevin Hall’s team at the US National Institutes of Health admitted 20 weight-stable adults to the NIH Clinical Center. Each person ate an ultra-processed diet for two weeks and an unprocessed diet for two weeks, in random order. The meals offered were matched for calories, energy density, macronutrients, sugar, sodium and fibre. People could eat as much or as little as they wanted.

• On the ultra-processed diet, people ate about 508 more calories a day, mostly from carbohydrate and fat.
• They gained about 0.9 kg over those two weeks, and lost about 0.9 kg on the unprocessed diet.

It is a randomised trial, which is why it matters. It is also 20 people over four weeks in a hospital ward, which is why it cannot settle long-term questions on its own.

The big picture: many studies, mostly observational

A 2024 umbrella review in The BMJ (Lane and colleagues) brought together 45 pooled analyses from published meta-analyses, covering almost 9.9 million people. Higher ultra-processed food intake showed a direct association with worse outcomes in 32 of those 45 pooled analyses, including cardiovascular deaths, type 2 diabetes, anxiety and common mental disorders, all-cause mortality and obesity.

The limits matter:

• These are associations, not proof of cause. People who eat a lot of ultra-processed food also differ in income, time, stress and much else.
• The authors’ own grading is cautious. Of the pooled analyses, 22 were rated low quality, 19 very low and only 4 moderate under GRADE.
• The review says the findings support more mechanistic research. It does not say the cause is fully understood.

The honest criticism of NOVA

Because NOVA classifies by processing, it puts very different products in one group. The FAO report’s own group 4 examples include carbonated soft drinks, fruit yoghurt, mass-produced packaged breads and baby formula. Those are very different products to put in one bucket. The US agencies’ request for information, covered below, even asked whether “ultra-processed” is the best term to use. A sensible reading: group 4 is a flag worth noticing, not a moral category, and the share of your eating it takes up matters more than any single item.

Where the official definition stands

On 25 July 2025 the US Food and Drug Administration and the Department of Agriculture published a joint Request for Information in the Federal Register (docket FDA-2025-N-1793). It asked for data to develop “a uniform definition of ultra-processed foods” for the US food supply. On 19 September 2025 they extended the comment period to 23 October 2025.

When I queried the Federal Register on 27 October 2026 for documents mentioning “ultra-processed” published since August 2025, the only notice about the definition itself was that extension. The other hits were unrelated notices that use the word in passing. No federal definition has been published there.

The definition has moved, though. The White House Office of Management and Budget’s review list shows an FDA item titled “White Paper: Proposed Definition of Ultra-Processed Food” (RIN 0910-ZD60), received on 3 August 2026. When I checked on 28 October 2026 its review was still listed as pending, and I did not find the text published. Until it is published, any “ultra-processed” claim on a US pack or in an ad has no single official meaning. NOVA is a widely used yardstick in the research.

A three-step label check you can do in the shop

1. Skip the front, flip the box. Claims like “whole grain”, “natural” or “high protein” describe what is in the product. They do not tell you how it was made.
2. Scan for one marker. In the first half of the list, look for kitchen-strange substances: syrups, maltodextrin, dextrose, isolates, hydrogenated oil. In the second half, look for flavour, colour, emulsifier, sweetener or gum. One is enough to put it in group 4.
3. Look for a group 1 or 3 swap, not perfection. Plain oats plus your own fruit, bread with four ingredients, plain yogurt you flavour yourself. The aim is to move the default for the things you eat most often, not to inspect every snack.

None of this needs a gadget. If you already track your metabolism, or are curious about it, these reviews cover the optional data side and what the evidence does and does not support:

• our guide to continuous glucose monitoring;
• our Levels review, which covers cost and what the evidence supports;
• the Zero app review, if meal timing is the lever you are working on.

If you would rather have a structured, paid programme than an app, Eat Stop Eat is Brad Pilon’s intermittent-fasting method: a longer fast once or twice a week, normal meals the rest of the time. We have not tested it ourselves. Its sales page says it can reverse signs of aging, ease diabetes symptoms and help joints heal; the fasting research does not establish those claims, so read it with the same eye you now bring to an ingredient list. Skip fasting programmes altogether if you are under 18, are underweight, have diabetes, take glucose-lowering medication, are pregnant or breastfeeding, or have a history of disordered eating, unless your clinician agrees.

See Eat Stop Eat

For the whole health room in one place, start with Health Unhacked: The Definitive Manual.

Disclaimer: this article explains a food classification system and summarises published research. It does not diagnose, treat or prevent any condition, and no food or dietary change described here is a cure for anything. Individual needs differ, especially for children, pregnancy, older adults and anyone with a medical condition. Ask a qualified clinician or registered dietitian before making significant changes.

Sources and how we checked

The main facts in this article were checked against these primary sources on 27 October 2026, and the OMB review status on 28 October 2026. Food labels change; check the pack in your hand.

• Ultra-processed foods, diet quality, and health using the NOVA classification system (Monteiro, Cannon, Lawrence, Louzada, Machado; FAO, Rome 2019; ca5644en), pp. 6-12 (Food and Agriculture Organization of the United Nations)
• Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake (Hall et al.; Cell Metab 2019;30(1):67-77.e3; PMID 31105044) (NIH / PubMed)
• Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses (Lane et al.; BMJ 2024;384:e077310; PMID 38418082) (The BMJ / PubMed)
• Ultra-Processed Foods; Request for Information (FDA and USDA; Federal Register 2025-14089, 25 July 2025; docket FDA-2025-N-1793) (US Federal Register)
• Ultra-Processed Foods; Request for Information; Extension of Comment Period (Federal Register 2025-18169, 19 September 2025) (US Federal Register)
• Pending EO 12866 Regulatory Review: White Paper: Proposed Definition of Ultra-Processed Food (HHS/FDA; RIN 0910-ZD60; received 3 August 2026) (Office of Management and Budget, OIRA)
• USDA FoodData Central branded food records 2281790, 2503524, 2147584 and 2055432, retrieved 27 October 2026 through the public FoodData Central API (api.nal.usda.gov). Search fdc.nal.usda.gov for an ID to see the record. (US Department of Agriculture)