Ultra-Processed Foods: What Processing Changes About Food

Fact-checked: 3 November 2026.

You’re looking at two loaves of bread. One advertises whole grains and fibre; the other has a shorter ingredients list. Which is the better choice? The debate about ultra-processed foods, or UPFs, can make such decisions harder than they already are. One day only sugar seems to matter; the next, the packaging itself looks suspicious.

Neither approach gets you very far. Processing deserves attention alongside nutrients, the type of food, and the role a product plays in your diet. Research published in October 2026 gives us reasons to look more closely. It does not provide a scientific basis for dividing an entire cupboard into “good” and “toxic”.

The key points

• UPF is a processing category, not a complete nutritional verdict or a rule about the number of ingredients.
• Different studies answer different questions: how much do people eat, how does the body respond to a meal, and what is associated over time?
• The new findings add to nutrition research. They do not make calories, fibre, sugar or salt irrelevant.
• Comparing products that serve a similar purpose is more useful than insisting that everything must be homemade.

What “ultra-processed” actually means

The widely used NOVA classification considers the extent and purpose of processing. Its four groups range from foods changed very little to industrial formulations, for example using protein isolates, modified starches or cosmetic additives for taste and texture. The NOVA authors explain identification using ingredients and production methods.

The four groups are also described in the DGE’s discussion of UPF.

Classification depends on the actual product. “Bread” or “plant-based” isn’t enough information by itself. Researchers face difficulties too: the German Nutrition Society describes UPF as a contested concept without a uniform definition. That does not make NOVA useless for research. It does help explain why a shopping app’s verdict cannot replace advice about an individual’s diet.

Your own recipe offers a useful check on ingredient-counting rules. A stew with twelve ingredients does not automatically become a UPF because of the number twelve. Equally, a short list alone cannot tell you whether a product is a good fit for your diet. List length is a convenient shortcut, which is also what makes it easy to misuse.

What the experiments show

2019: Two menus, different amounts eaten

In the NIH trial by Hall and colleagues, 20 adults received a UPF-rich and an unprocessed diet for two weeks each, in random order. They could eat according to appetite. During the UPF phase, they consumed about 508 additional kilocalories per day on average and gained roughly 0.9 kilograms; during the other phase they lost roughly 0.9 kilograms. This establishes an effect of those menus under controlled conditions. It does not isolate the particular food property responsible.

That distinction matters. A whole meal contains more than one experimental variable. Change breakfast, lunch and dinner, and you have changed a dietary pattern. You cannot immediately work backwards from the result to identify which additive in which product contributed how much.

2025: The quality of the whole menu matters too

The UPDATE trial randomised 55 adults with overweight or obesity to two eight-week dietary phases; 50 contributed data to the primary analysis. Both menus met UK dietary guidelines. Weight fell in both phases, but more with minimally processed food: 2.06 versus 1.05 per cent. A small, selected group and limited follow-up cannot yield a personal weight-loss forecast. The finding also illustrates why “UPF” and “nutritional quality” are not interchangeable descriptions.

October 2026: A meal produces different metabolic responses

The 5 October study by Hutelin and colleagues included 57 adults; 32 took part in metabolic testing. In a randomised crossover comparison, they ate nutritionally matched UPF and non-UPF meals. Insulin response and energy expenditure were higher after UPF, while carbohydrate oxidation was lower. Associations with brain responses to food cues were also found. These were acute responses, not later disease outcomes. Different protein sources and incompletely measured food structures limit the explanation.

A crossover design has a useful strength: participants experience both conditions, acting as their own comparison. This reduces the influence of stable differences between people. It does not turn two different meals into an experiment in which only one processing step has changed.

Another distinction helps when interpreting the findings. A measurable difference can be scientifically interesting without already being a reliable warning sign for everyday life. Turning a metabolic curve directly into a prediction of disease skips precisely the research needed to make that connection.

Why texture and eating speed belong in the explanation

“Processing” bundles together many properties. A randomised texture experiment involving 50 people compared softer and harder meals in both processing categories. Harder textures were eaten more slowly and were associated with lower intake. That supports investigating specific properties of foods rather than looking for the entire explanation in a category name.

What the new study in children can add

A JAMA cohort study published on 23 October followed 46 children of healthy weight at baseline. Higher habitual UPF intake was associated with stronger dorsal striatal responses to food images; those brain responses were associated with a rise in body fat percentage over a year. Diet was not randomly assigned. This is a clue about possible relationships, not proof of addiction or food-induced brain damage.

Observational studies complement experiments; they do not replace them. Dietary reports, living conditions and other behaviours may be linked. The UK Scientific Advisory Committee on Nutrition’s evidence assessment identifies issues including residual confounding and classification difficulties. “Associated with” is therefore the accurate wording for such findings, not a rhetorical attempt to play them down.

A shopping approach without a list of bans

The German Nutrition Society’s recommendations remain a useful starting point: predominantly plant-based foods, including vegetables, fruit, whole grains, pulses and nuts, with water as the default drink. These general recommendations are for healthy adults aged 18–65 who eat a mixed diet. The comparisons below are practical examples, not a clinically tested UPF treatment.

1. Compare products that do the same job. With two breakfast products, you can consider ingredients, fibre and sugar together. A shorter list does not automatically justify replacing a whole-grain product with a lower-fibre alternative. With two loaves, start by considering what you eat the bread with and how often, rather than which packet appears morally spotless.

2. Plan the alternative as well. “Fewer ready meals” isn’t yet a dinner. One example might be keeping frozen vegetables, tinned beans and a staple you enjoy on hand. Whether that works for you also depends on price, time, kitchen equipment and taste. A plan that requires you to live a different life every evening is not a particularly good plan.

3. Start with what you actually buy often. Someone who regularly drinks lemonade could try water as an alternative. Someone who hardly ever drinks sugary drinks does not need a universal “drinks first” programme. Use an ordinary shop as your starting point: which two products keep appearing, and what realistic replacement would you willingly buy again?

4. Assess the swap, not just the removal. A protein bar might replace another food or be eaten on top of it. A sauce might make a meal possible that you otherwise wouldn’t cook. A useful decision takes that role into account. For the separate question of protein needs, see our guide to the protein hype.

Keep your experiment small enough that you would still want to do it after a normal working day. You do not have to classify every gram or calculate a daily UPF allowance. Start by observing something simpler: do you enjoy the alternative, does it fit your budget, and will it be on the next shopping list? This does not promise a particular change in weight or health. It is a way to turn an abstract debate into a practical decision you can assess for yourself.

Health note: This article does not replace medical advice or individual nutritional therapy. If you have an eating disorder, relevant medical conditions or a prescribed diet, discuss dietary changes with your treating professional.

Sources and scope

This selection combines experiments, an observational study and professional guidance. It is not a systematic review of all UPF research. Checked on 3 November 2026:

• Monteiro et al., 2019: NOVA and identifying UPFs – the classification.
• Hall et al., 2019: the controlled NIH feeding trial – energy intake and weight.
• Dicken et al., 2025: the UPDATE trial – two menus meeting dietary guidelines.
• Hutelin et al., 2026: metabolic and neural responses – an acute meal comparison.
• Teo et al., 2022: texture and eating rate – properties of the meal.
• Alves et al., 2026: the child cohort – associations over one year.
• SACN’s 2025 evidence assessment, the DGE’s discussion of UPF and DGE dietary guidance – limitations and everyday orientation.