How to spot ultra-processed food from the label

The NOVA classification

Developed at the University of São Paulo, NOVA sorts food by the extent and purpose of processing rather than by nutrient content. It has become the standard framework in nutrition research.

Group 1: unprocessed and minimally processed. Fruit, vegetables, meat, fish, eggs, milk, grains, pulses. Processing removes inedible parts or preserves: washing, drying, freezing, pasteurising.

Group 2: processed culinary ingredients. Oil, butter, sugar, salt, flour. Extracted from group 1 and used to cook with, not eaten alone.

Group 3: processed foods. Group 1 plus group 2. Bread, cheese, tinned vegetables, cured meat, canned fish. Recognisably made from food.

Group 4: ultra-processed. Industrial formulations, typically five or more ingredients, containing substances not used in domestic kitchens, plus additives whose purpose is to make the result palatable and shelf-stable.

The distinction people get wrong

Processed is not ultra-processed. Tinned tomatoes are processed. Wholemeal bread is processed. Cheese is processed. Processing is one of the more important technologies humans have. It is why food is safe and available out of season.

Ultra-processing is a different thing: formulation from industrial ingredients rather than transformation of food. That is the distinction NOVA is drawing, and losing it turns a useful framework into a slogan.

The marker ingredients

Certain ingredients appear almost exclusively in group 4. Spot any of these and you are almost certainly holding UPF:

Never used in home cooking:

• Protein isolate, protein concentrate, hydrolysed protein
• Modified starch
• Invert sugar, glucose-fructose syrup, high-fructose corn syrup, maltodextrin
• Hydrogenated or interesterified oil
• Soy protein isolate, milk protein concentrate

Cosmetic additives: there to make an industrial formulation look and taste like food:

• Colours (E100–E199)
• Flavourings and flavour enhancers (E620–E650)
• Emulsifiers (E322, E471, E472, E433, E466)
• Thickeners and stabilisers
• Non-nutritive sweeteners (E950–E969)
• Bulking agents, firming agents, anti-foaming agents

What the evidence says

Large cohort studies consistently associate high UPF consumption with higher risk of obesity, cardiovascular disease, type 2 diabetes and all-cause mortality. A landmark NIH randomised controlled trial found that people ate around 500 kcal more per day on an ultra-processed diet than on a minimally processed one matched for calories, sugar, fat, fibre and macronutrients, and gained weight accordingly.

That last detail is the interesting one. It suggests something about ultra-processing itself, whether that is energy density, eating speed, texture, palatability or effects on satiety signalling, rather than simply the nutrient profile.

The mechanism is not settled, and the category is broad: wholemeal bread with an emulsifier and a bag of sweets are both group 4, and treating them as equivalent is clearly wrong. Critics of NOVA make this point fairly.

Practical use

Do not try to eliminate group 4. That is neither realistic nor necessary, and pursuing it tends to produce anxiety rather than health.

Use it as a gradient. Between two similar products, the one with the shorter, more recognisable ingredient list is usually the better choice, and that is a judgement you can make in three seconds without any nutritional expertise.

The count of additives on a label is a decent proxy. A food needing six colours, three emulsifiers and two flavour enhancers is telling you how far it sits from its ingredients.

Where scanning fits

A scan returns the full ingredient list with additives named and counted, which makes the marker ingredients visible immediately instead of requiring you to parse a paragraph of small type. Comparing two products in the same category then shows you which is the shorter, simpler formulation, which is, in the end, most of what NOVA is asking you to notice.