What Counts as Ultra-Processed Food, and How Much It Matters

“Ultra-processed” has become one of those words that turns up constantly in food headlines. It gets attached to studies on heart disease and depression, and to long arguments about whether sliced bread counts as junk food. The concern behind it is reasonable. The label, though, covers far more ground than most people realise, and that makes the research harder to read than the headlines suggest.

The term comes from NOVA, a system developed by Carlos Monteiro’s group at the University of São Paulo. It sorts food by how much it has been processed rather than by how much fat or sugar it contains. Once you know where it draws its lines, both the studies and the supermarket shelves make a bit more sense.

What NOVA Counts as Ultra-Processed

NOVA has four groups. The first is food that is unprocessed or barely processed: fresh or frozen vegetables, eggs, milk, plain meat, dried beans. The second is culinary ingredients such as oil, butter, sugar and salt. The third is processed food made by combining the first two in fairly simple ways, like cheese, tinned fish, or bread made from flour, water, yeast and salt.

The fourth group is different in kind. Monteiro describes ultra-processed foods as industrial formulations made mostly from substances extracted from foods or synthesised, put together with additives such as flavourings, colours, emulsifiers and sweeteners. Soft drinks, packaged snacks, sweets, many breakfast cereals and reconstituted meat products all land here. The quickest check is the ingredient list. If it’s long and includes things you would never keep in a kitchen, such as dextrose, invert sugar syrup or mono- and diglycerides of fatty acids, you’re probably holding one.

Monteiro, C.A. et al. (2019). “Ultra-processed foods: what they are and how to identify them.” Public Health Nutrition, 22(5), 936-941.

The Trial Everyone Cites

Most of what we know about ultra-processed food comes from watching what large groups of people eat. The best-known controlled test is a small one. In 2019, Kevin Hall and colleagues at the US National Institutes of Health had 20 adults live on a metabolic ward for four weeks, where everything they ate was measured. They spent two weeks on an ultra-processed diet and two weeks on a minimally processed one. The meals on offer were matched for calories, sugar, fat, fibre and macronutrients, and people could eat as much or as little as they liked.

On the ultra-processed diet, they ate about 500 kcal a day more and ate faster. They gained about 0.9 kg over those two weeks, and lost about 0.9 kg on the minimally processed diet. Since the menus were matched on paper, the result suggested that something about the way ultra-processed food is made, and not only its sugar or fat, leads people to eat more of it. Exactly what that is remains an open question, and eating speed is one candidate. Twenty people and two weeks per diet is also a long way from showing what happens over years.

Hall, K.D. et al. (2019). “Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake.” Cell Metabolism, 30(1), 67-77.

What the Large Reviews Show

The observational evidence is large. A 2024 umbrella review in the BMJ brought together 45 pooled analyses covering about 10 million people. Higher intake of ultra-processed food was associated with 32 harmful health outcomes. The most convincing links were with death from cardiovascular disease, type 2 diabetes, and common mental disorders such as anxiety and depression.

Lane, M.M. et al. (2024). “Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses.” BMJ, 384, e077310.

These are associations, and the category makes them harder to interpret. NOVA puts a packaged wholegrain loaf that contains an emulsifier in the same group as a two-litre bottle of soft drink. People who eat a lot of ultra-processed food can also differ in income and in how much time they have to cook, and researchers can only adjust for so much. The averages can’t tell you which foods inside the group drive the risk, and it seems unlikely that all of them do so equally.

A Practical Way to Shop

None of this means you should treat everything in a packet with suspicion. Tinned tomatoes, frozen peas, plain yoghurt and oats aren’t ultra-processed at all. The more useful targets are the ultra-processed foods that stand in for a meal or a drink: soft drinks, packaged snacks, processed meat and sweets. Replacing one of those with something you cook, even something simple, is probably a better use of effort than agonising over which bread to buy.

The ingredient list on the back tells you more than the claims on the front. “High protein” or “made with real fruit” says nothing about how a product was made. Cooking a bit more often helps from the other side. A pot of soup on Sunday, or oats soaked the night before, covers several of the meals that would otherwise come out of a wrapper.

If you want a quick read on the packaged food you already buy, Lamplit’s barcode scanner is free and has no monthly limit. Scan a product and it shows a health score, the Nutri-Score and the NOVA group when Open Food Facts has that data, along with any additives listed. This week, try scanning the five packaged things you eat most often. The results may not match your guesses, and they give you a short list to work on instead of a whole supermarket to worry about.

Scan the first item from your cupboard in Lamplit and see its NOVA group. Barcode scanning is free.