How to tell if a food is ultra-processed

The test in one sentence

NOVA, the classification behind every ultra-processed food study cited on this site, sorts food by what was done to it rather than by its nutrition. Group 1 is unprocessed or minimally processed. Group 2 is culinary ingredients such as oil, butter, sugar and salt. Group 3 is the two combined into products such as cheese, tinned vegetables and fresh bread. Group 4 is everything containing at least one substance that exists only as an industrial input.

That last clause is the whole test. Not "is it processed", which describes bread and cheese and yoghurt, but "does it contain something no kitchen has".

The marker ingredients

These are the ones that move a product into group 4 on their own:

• Protein isolates and hydrolysates, including whey, soy and pea protein isolate
• Modified starch, maltodextrin, glucose syrup, glucose-fructose syrup, invert sugar
• Hydrogenated, interesterified or fractionated oils
• Emulsifiers: lecithins, mono- and diglycerides, polysorbates, carboxymethylcellulose
• Thickeners and stabilisers: carrageenan, xanthan, guar and gellan gum
• Flavourings: "natural flavour", "artificial flavour", flavour enhancers
• Colours, whether synthetic or extracted
• Non-sugar sweeteners: aspartame, sucralose, acesulfame K, steviol glycosides
• Bulking, anti-caking, firming, glazing and humectant agents

What the classification is not

NOVA is not a health score and it is not a safety judgement about any single additive. A wholemeal loaf with an emulsifier in it is group 4. So is a sugar-free drink. The classification is describing an industrial process, and its usefulness comes from the fact that the process, rather than any one ingredient, is what the research tracks.

It also has a real weakness worth knowing: it puts a supermarket sandwich and a can of cola in the same group. The category is broad, and treating everything in it as equivalent is not supported.

Why it is worth the trouble

Most nutrition findings are observational, which means the healthier group might simply be healthier people. Ultra-processing is the rare case with a proper randomised trial in humans.

In 2019 a team at the US National Institutes of Health housed twenty adults and fed them ultra-processed and unprocessed diets for two weeks each, matched for the calories presented and for sugar, fat, sodium and fibre. Same numbers on the label. People ate about 500 kcal a day more on the ultra-processed diet and gained weight.

That isolates the processing itself. It was not the sugar and it was not the fat, because those were matched. Something about the form of the food drove the overeating. It is the strongest single result in this field and it is the reason the classification is worth applying at the shelf.

Worked examples

Real products, scored by the app itself, with every reason the engine gave.

• Froot LoopsKellogg's33Poor
• Beanz in a rich tomato sauceHeinz49Poor

Common questions

Is all processed food ultra-processed?

No. Tinned tomatoes, plain yoghurt, cheese, dried pasta and frozen vegetables are processed and are not group 4. The distinction is industrial-only ingredients, not processing as such.

Does a long ingredient list mean ultra-processed?

Not by itself, though the two correlate. One marker ingredient decides it. Optimally applies an additional deduction past fifteen ingredients, separately from the cap.