Ultra-processed foods are industrial formulations, such as soft drinks, packaged snacks and reconstituted meats, defined by the NOVA system. A 2024 BMJ umbrella review linked higher intake to 32 adverse health outcomes, but most evidence is observational. Two controlled feeding trials, in 2019 and 2025, found these diets led to more eating or less weight loss.
What makes a food "ultra-processed"
Almost everything we eat is processed in some way: washed, chopped, frozen, cooked, fermented or canned. "Ultra-processed" means something narrower. The term comes from NOVA, a classification developed by researchers at the University of São Paulo in Brazil, which sorts foods by "the extent and purpose of industrial processing" rather than by their nutrients (Monteiro and colleagues, Public Health Nutrition 2019).
NOVA has four groups. A 2019 report the NOVA authors prepared for the Food and Agriculture Organization of the United Nations (FAO) gives the definitions and examples (FAO, Ultra-processed foods, diet quality, and health using the NOVA classification system).
| NOVA group | What it is | Examples from the FAO report |
|---|---|---|
| 1. Unprocessed or minimally processed | Whole foods, perhaps dried, frozen, ground, pasteurised or fermented, with nothing added | Fresh or frozen fruit and vegetables, rice, beans and lentils, meat and fish, eggs, milk, plain yoghurt, unsalted nuts, oats, pasta made from flour and water |
| 2. Processed culinary ingredients | Substances pressed, refined or extracted from group 1 foods, used in cooking | Vegetable oils, butter, lard, sugar, honey, salt |
| 3. Processed foods | Group 1 foods with salt, sugar or oil added, preserved by canning, bottling or fermentation | Canned vegetables and fish, salted nuts, cured meats, fruit in syrup, freshly made unpackaged breads and cheeses |
| 4. Ultra-processed foods | Formulations "mostly of exclusive industrial use", made by a series of industrial processes | Soft drinks, packaged snacks, confectionery, ice cream, mass-produced packaged breads, margarines, breakfast cereals, fruit yoghurts, nuggets, sausages, instant soups and noodles |
The NOVA authors offer a practical test. A product is ultra-processed if its ingredients list contains at least one item "characteristic of the NOVA ultra-processed food group": either food substances "never or rarely used in kitchens", such as high-fructose corn syrup, hydrogenated or interesterified oils and hydrolysed proteins, or additives used to make the product more palatable or appealing, such as flavours, flavour enhancers, colours, emulsifiers, sweeteners and thickeners.
That test explains some results that surprise people. A loaf of bread made from flour, water, yeast and salt is group 3; a mass-produced loaf with emulsifiers is group 4, even if it is wholemeal. Plain yoghurt is group 1; a fruit yoghurt with flavourings and thickeners is group 4.
Reading a label: three worked examples
Applying the NOVA test in a supermarket takes practice, because two products with the same name can sit in different groups. These examples use the ingredient rule above.
Baked beans. A tin listing beans, tomatoes, water, sugar, salt and spices is a processed food (group 3): group 1 foods with group 2 ingredients added and preserved by canning. If the label adds modified starch, flavourings or a sweetener such as sucralose, the same product becomes group 4. Either way, beans in tomato sauce supply fibre and protein, so the practical advice is to pick a lower-sugar, lower-salt version rather than to give them up.
Porridge oats. Plain rolled oats are group 1. A sachet of "golden syrup" instant oats with flavourings, an emulsifier and added sugar is group 4. Making porridge from plain oats with milk and adding your own fruit takes only a few minutes more than the sachet.
Chicken. A raw chicken breast is group 1. Smoked chicken cured with salt is group 3, because NOVA puts salted, cured and smoked meats there. Breaded nuggets made from reconstituted meat with dextrose, starches and flavourings are group 4.
The pattern is the same each time: the number of industrial ingredients you would not keep in your own kitchen is the clue. That is why the NOVA authors suggest reading the ingredients list rather than the nutrition panel when classifying a product, and why many dietitians suggest using both.
How much ultra-processed food people eat
In the United States, ultra-processed foods make up most of the diet. The CDC's National Center for Health Statistics reported that during August 2021 to August 2023, the mean share of calories from ultra-processed foods was 55.0% among everyone aged 1 and over, with young people aged 1 to 18 getting 61.9% and adults 53.0% (NCHS Data Brief No. 536, August 2025). Sandwiches including burgers, sweet bakery products, savoury snacks and sweetened drinks were four of the top five sources for both age groups.
These figures sit behind much of the current US policy debate; our United States country guide covers the wider nutrition landscape there.
Top 10 ultra-processed foods to recognise
There is no official "top 10", but combining the NCHS list of the biggest calorie sources with NOVA's group 4 examples gives a useful picture of the products that most often fill a shopping basket:
- Sugar-sweetened soft drinks and energy drinks
- Packaged sweet bakery products: biscuits, cakes, pastries and doughnuts
- Savoury packaged snacks such as crisps and flavoured crackers
- Confectionery and chocolate bars
- Ice cream and dairy desserts with added flavourings
- Mass-produced packaged breads, buns and tortillas with additives
- Sweetened breakfast cereals and cereal bars
- Reconstituted meat products: sausages, hot dogs, nuggets and many burgers
- Ready meals, pizzas and instant noodles or soups
- Flavoured yoghurts and milk drinks
Some of these, such as wholegrain breakfast cereals or packaged wholemeal bread, can still supply useful fibre. That is the heart of the debate below.
What the evidence shows
The observational evidence
Most research on ultra-processed food comes from cohort studies, which follow large groups of people and compare those who eat more with those who eat less. A 2024 umbrella review in The BMJ pooled 45 analyses covering almost 9.9 million participants and found direct associations between higher ultra-processed food intake and 32 health outcomes, including mortality, cancer, and mental, respiratory, cardiovascular, gastrointestinal and metabolic health (Lane and colleagues, BMJ 2024).
The review graded the evidence carefully. It rated as "convincing" the associations with cardiovascular death, type 2 diabetes and common mental disorders, and as "highly suggestive" those with all-cause death, heart disease death, obesity, depression and sleep problems. But the certainty using the GRADE framework was mostly low or very low: only four analyses were moderate quality. Cohort studies can show that two things go together, not that one causes the other, and people who eat more ultra-processed food often differ in income, smoking, activity and other ways that are hard to adjust for.
The controlled trials
Two feeding trials have tested the effect directly.
- NIH inpatient trial, 2019. Twenty adults lived at the US National Institutes of Health Clinical Center for four weeks and ate an ultra-processed diet for two weeks and an unprocessed diet for two weeks, in random order. The diets offered matched calories, sugar, salt, fibre and macronutrients, and people ate as much as they liked. On the ultra-processed diet they ate about 508 kcal a day more and gained 0.9 kg; on the unprocessed diet they lost 0.9 kg (Hall and colleagues, Cell Metabolism 2019).
- UPDATE trial in England, 2025. Fifty-five adults with a body mass index of 25 or more ate two eight-week diets that both followed the UK Eatwell Guide, one minimally processed and one ultra-processed, in random order. Both diets led to weight loss, but weight fell by 2.06% on the minimally processed diet against 1.05% on the ultra-processed one, a significant difference (Dicken and colleagues, Nature Medicine 2025).
These trials are small and short, but they are the strongest evidence that processing itself, not just sugar, salt and fat, can affect how much people eat. The UPDATE result also matters for a different reason: an ultra-processed diet that met national guidelines still led to weight loss, just less of it.
Why it might matter
Researchers have proposed several mechanisms: high energy density, soft textures that are eaten quickly, combinations of fat, sugar and salt that make foods "hyper-palatable", loss of the natural food structure, and particular additives. In November 2025 The Lancet published a three-part series led by the NOVA team, which concluded that "displacement of long-established dietary patterns by ultra-processed foods is a key driver" of rising diet-related chronic disease (Monteiro and colleagues, Lancet 2025). That is a strong position from the classification's own authors; other scientists, described below, think the evidence is not yet that clear.
Is the NOVA category too broad?
The sharpest scientific criticism is not that ultra-processed food is healthy, but that the category lumps very different foods together.
- Wholegrain bread and yoghurt. In three large US cohorts, total ultra-processed food intake was linked to a higher risk of type 2 diabetes, but not every subgroup behaved the same way. Refined breads, sauces, sweetened and artificially sweetened drinks, animal-based products and ready meals were linked to higher risk, while cereals, dark and wholegrain breads, fruit-based products, and yoghurt and dairy desserts were linked to lower risk (Chen and colleagues, Diabetes Care 2023).
- Experts disagree on which foods belong where. When 159 to 177 French food and nutrition specialists were asked to assign foods to NOVA groups, agreement was low, even when they had the ingredients list, leading the authors to conclude that NOVA criteria "do not allow for robust and functional food assignments" (Braesco and colleagues, European Journal of Clinical Nutrition 2022).
- Overlap with existing advice. Nutrition scientist Michael Gibney has argued that the definition and the list of example foods have changed over time, and that across national surveys higher ultra-processed intake tracked sugars and fibre more than total fat, saturated fat or sodium (Gibney, Current Developments in Nutrition 2019).
The UK's Scientific Advisory Committee on Nutrition (SACN) reached a cautious middle position in its July 2023 statement. It accepted that higher consumption of processed foods "was associated with increased risks of adverse health outcomes", but said there were "uncertainties around the quality of evidence", that confounding "may not be adequately accounted for", and that the NOVA categories are "very broad and capture a wide range of foods". It said the evidence "needs to be treated with caution" and called for a classification system that can be applied reliably, plus better trials and cohort studies (SACN statement on processed foods and health, summary).
Ultra-processed is not the same as unhealthy, and minimally processed is not the same as healthy. Packaged wholemeal bread, plain frozen vegetables and some fortified foods can be useful, while a home-made cake is still a cake. Use processing as one lens, alongside fibre, sugar, salt and saturated fat.
Brazil's food guide and the policy response
Brazil built its national dietary advice around processing. Its Ministry of Health's 2014 Dietary Guidelines for the Brazilian Population, the second edition after a 2006 version, uses the degree of processing as its main criterion. In summary, it advises people to base their diet on natural or minimally processed foods, to use oils, fats, salt and sugar in small amounts for cooking, to limit processed foods, and to avoid ultra-processed foods (Georgetown Global Food Laws database). Our Brazil country guide explains how dietitians there apply it.
Other countries have since moved towards similar language without adopting NOVA in full:
- Canada. Canada's food guide asks people to limit highly processed foods, defined as "processed or prepared foods and drinks that add excess sodium, sugars or saturated fat to the diets of Canadians" (Health Canada). Notice that this definition is based on nutrients, not on industrial ingredients.
- United States. In July 2025 the FDA and USDA asked the public for data and information to help develop "a uniform definition of ultra-processed foods", with comments closing on 23 October 2025 after an extension (Federal Register, 25 July 2025). When we checked the Federal Register on 17 October 2026, no final federal definition had been published. Meanwhile, the January 2026 Dietary Guidelines tell Americans to avoid highly processed foods; our explainer on the new food pyramid covers that advice, and our guide to the MAHA report covers the wider political programme behind it.
- United Kingdom. The House of Lords Food, Diet and Obesity Committee published its report Recipe for health: a plan to fix our broken food system on 24 October 2024 (UK Parliament). For the UK professional landscape, see our United Kingdom country guide.
Policy also has to consider cost and time. The NOVA authors themselves describe ultra-processed products as designed to be convenient and ready to consume, so telling people to avoid them without addressing price, cooking time and access can widen health gaps. Our article on food insecurity looks at that problem, and our guide to food as medicine covers programmes that prescribe healthy food.
Practical ways to eat less ultra-processed food
You do not need to eliminate ultra-processed foods to benefit. Shifting the balance is realistic and fits every set of national guidelines.
- Start with drinks. Replacing sugary and sweetened drinks with water, milk or unsweetened tea and coffee is the easiest big change.
- Make breakfast simpler. Swap sweetened cereals for porridge oats, plain yoghurt with fruit, or eggs on wholemeal toast.
- Read the ingredients list, not just the front of the pack. If it includes several ingredients you would not find in a kitchen, it is probably group 4. Use this as a prompt to compare, not a rule.
- Cook in batches. A pot of soup, chilli or dhal covers several meals and replaces ready meals.
- Upgrade snacks. Fruit, unsalted nuts, plain popcorn and hummus with vegetables replace crisps and biscuits; our guide to high-protein snacks has more ideas.
- Keep useful convenience foods. Frozen vegetables, tinned beans and fish, and plain wholegrain bread are quick and nutritious.
- Prioritise fibre. Many ultra-processed foods are low in it; our fibre calculator helps you track it.
To see how your eating compares with an evidence-based pattern, try the free healthy eating check. If you are worried about fats as well as processing, our review of seed oils looks at one of the most argued-over ingredients. For a personal plan that fits your budget and routine, you can book a session with a registered dietitian.
Get the evidence-based nutrition newsletter
Frequently asked questions
What are the top 10 ultra-processed foods?
There is no official list, but the products that supply most calories include soft drinks, packaged cakes and biscuits, crisps and savoury snacks, confectionery, ice cream, mass-produced packaged breads, sweetened breakfast cereals, sausages and nuggets, ready meals and pizzas, and flavoured yoghurts. In the US, NCHS data for 2021 to 2023 put sandwiches and burgers at the top of the calorie list.
What is the no. 1 unhealthiest food?
No single food is the unhealthiest, because health depends on your whole diet and how much you eat. Sugar-sweetened drinks are a strong candidate for the most worth cutting: the 2025–2030 US Dietary Guidelines tell people to avoid them, and they add sugar without fibre or protein. Processed meats are another: WHO's cancer agency classifies processed meat as Group 1 because the evidence shows it causes colorectal cancer (WHO).
Is peanut butter an ultra-processed food?
It depends on the jar. NOVA puts plain nuts in group 1 and salted nuts in group 3, so peanut butter made from peanuts and a little salt is not ultra-processed. A spread with added sugars plus emulsifiers, hydrogenated oils or flavourings meets NOVA's ingredient test for ultra-processing. NOVA documents do not list peanut butter by name, so check the ingredients.
Is Greek yogurt ultra-processed?
Plain Greek yoghurt is not. The FAO report on NOVA lists plain yoghurt in group 1, minimally processed foods. Flavoured or fruit yoghurts that contain flavourings, sweeteners, thickeners or colours are classed as ultra-processed. Some cohort studies have linked yoghurt and dairy desserts with lower diabetes risk, showing why many researchers say the category needs refining.
Are all ultra-processed foods bad for you?
Not necessarily. Large cohort studies find total ultra-processed intake linked with poorer health, but some subgroups, such as wholegrain breads and breakfast cereals, are linked with lower diabetes risk. The UK's SACN says the evidence needs caution. A sensible approach is to cut sugary drinks, snacks and processed meats first, and keep nutritious convenience foods.
Is there an official definition of ultra-processed food?
Not in US federal rules as of October 2026, when the FDA and USDA had not yet published a final definition. Most research uses the NOVA classification from Brazil. The FDA and USDA asked for public input on a uniform US definition in July 2025, and Canada's food guide uses a nutrient-based description of highly processed foods instead. SACN has called for a classification that can be applied reliably.
