This is the top-ranked entry on the register by population burden, and it is also the one where the underlying category is weakest. Both of those are true at once, and a page that only says one of them is misleading.
1. What the observational evidence says
A 2024 umbrella review in the BMJ pooled 45 meta-analyses covering roughly 9.9 million people and reported associations between higher ultra-processed intake and 32 adverse health outcomes — including around 50% higher cardiovascular mortality, roughly 21% higher all-cause mortality, and higher risk of type 2 diabetes, several cancers, adverse mental health outcomes and obesity.
That is a lot of outcomes, which cuts both ways. A single exposure associated with thirty-two different diseases is either something very fundamental, or a marker for something else — poverty, overall diet quality, smoking, activity, or all of them together. Umbrella reviews inherit every weakness of the studies inside them, and the underlying evidence is almost entirely observational and reliant on food frequency questionnaires. Strong for the association; the causal question is section 2.
2. The one experiment
Hall and colleagues, 2019 — small, brief, and the only randomised evidence there is
- Twenty adults living on a metabolic ward for four weeks, crossing over between an ultra-processed diet and an unprocessed one for two weeks each. The two diets were matched for calories presented, sugar, fat, fibre and macronutrients — the participants simply ate as much or as little as they wanted.
- On the ultra-processed diet they ate about 500 kcal a day more, and gained weight. On the unprocessed diet they lost it. Same offered calories. Same macros. Different outcome.
- Why it matters more than its size suggests: it is the only study that removes the confounding. Everyone in it was their own control, the food was weighed, and nobody was choosing between a ready meal and cooking from scratch on a Tuesday night after work.
- And why it does not settle it: twenty people, two weeks per arm, an artificial setting, a soft mechanism — the leading explanations are eating rate, energy density and texture rather than anything about additives — and it has not been replicated at scale.
The most defensible reading is that processing changes how much you eat, which is a narrower claim than the one usually made and is enough on its own to justify the advice. Claims that emulsifiers or preservatives directly cause disease are Emerging and rest largely on animal and laboratory work.
3. The problem with the category itself
“Ultra-processed” comes from the NOVA classification, which sorts food by how it was made rather than by what is in it. That is a genuinely useful idea and a genuinely blunt instrument.
| All ultra-processed | And yet |
|---|---|
| A can of cola A packet of crisps A supermarket ready meal |
Wholemeal supermarket bread Baked beans Plain low-sugar yoghurt Tinned soup Most infant formula Many meat substitutes |
- The right-hand column contains staples of a perfectly good diet, several of them cheap sources of fibre and protein. A rule that treats wholemeal bread and cola as the same thing is not a rule anyone should follow literally.
- The classification is also poorly reproducible — independent coders assigning the same foods disagree more often than a scientific category should allow, which weakens every study built on it.
- And it may simply be tracking older, better-understood things. Ultra-processed food is on average higher in free sugar, salt and saturated fat, lower in fibre, more energy-dense and cheaper — and diets high in it are commoner in poorer households. Disentangling “processing” from all of that is genuinely hard — and UK advisory opinion says so. The Scientific Advisory Committee on Nutrition's position statement on processed foods and health (July 2023), and its rapid evidence update (2025), describe the observed associations as concerning while setting out the limitations of the NOVA classification and the likelihood of residual confounding — and make no ultra-processed-specific dietary recommendation. SACN's own critique of the metabolic ward trial matches the caveats in section 2 above.
- None of which makes the association go away. It survives adjustment for nutritional composition in many analyses, which is why the idea has not been dismissed.
Why this page still ranks it first. The register orders by estimated population burden, not by how clean the evidence is. Ultra-processed food supplies well over half of average energy intake in the UK, the association is large and consistent, and there is experimental support for at least the mechanism that matters most. An entry can be top of the list and still have a contested definition, and saying so is more useful than pretending otherwise.
4. What to do with an imperfect category
The way out is to stop using the category as a rule and use the underlying properties instead. These are what the evidence actually points at, and they sort the two columns above correctly.
| Ask | Because |
|---|---|
| How fast can this be eaten? | Eating rate is a leading candidate mechanism. Food that is soft, dry and requires little chewing is eaten faster than satiety signalling can keep up with |
| How many calories per mouthful? | Energy density does most of the work in the Hall trial's likely explanation |
| Does it have fibre and protein in it? | The two things that produce fullness. This is what separates baked beans from crisps, and NOVA does not |
| Is it a drink with calories in it? | Liquid calories barely register as food at all. The single clearest sub-category |
| Would you stop after a normal portion? | The plain version of all of the above, and the one people can actually apply in a shop |
Cut the drinks with calories first
Clearest evidence, easiest change, no cooking skill required.
Then the snacks engineered for volume
Crisps, biscuits, confectionery, pastries. Fast to eat, energy-dense, low fibre — every box ticked.
Then displace, rather than restrict
Adding protein and fibre reduces intake more sustainably than removing things does. See building the plate and the five things worth doing, where the same argument is made about satiety beating restriction.
Keep the useful processed foods
Tinned pulses and fish, frozen vegetables, plain yoghurt, wholemeal bread, porridge oats. Processing is not the enemy; a category is not a diet. Being poorer and busier is the commonest reason for a high ultra-processed share, and advice that ignores that is advice for somebody else.
Preservatives and additives specifically are covered in food and product safety, which is the topic for what is in the food; this page is about what the processing does to how much of it you eat.