What is an ultra-processed food?

It is a food classed in group 4 of the NOVA classification: an industrial formulation made mainly from substances extracted or derived from food, with cosmetic additives such as flavourings, colourings, emulsifiers or sweeteners. It is not a nutritional definition: it is based on processes and ingredients.

How to spot an ultra-processed food on the label?

Look in the ingredient list for things you would not find in a home kitchen: glucose-fructose syrup, maltodextrin, isolated or hydrolysed proteins, hydrogenated oils, flavourings, emulsifiers, colourings, sweeteners. A long list with technical names or E-numbers is another clue. A single one of these markers is enough to classify a product as NOVA 4.

Are ultra-processed foods proven to be dangerous?

Observational studies associate high consumption with a higher risk of mortality and chronic disease, but ANSES considers the weight of evidence to be low, and the association does not by itself establish a cause. Kevin Hall's trial (2019) shows that on an ultra-processed diet, people consume around 508 kcal more per day, over two weeks and with only 20 adults.

Are all NOVA 4 products bad?

No. NOVA measures processing, not nutritional quality. Some NOVA 4 products have a decent nutritional profile, such as sliced bread or yoghurts rated well on the Nutri-Score, and some non-ultra-processed foods are poorly balanced, such as a homemade cake. The most useful approach is to cross-check NOVA, the Nutri-Score and frequency of consumption, aiming for gradual reduction.

Ultra-processed foods: how to recognise them and which to avoid

What does "ultra-processed" mean?

The term comes from theNOVAclassification, created by a team of Brazilian researchers led by Carlos Monteiro. It does not classify foods by their sugar or fat content, but by the extent and purpose of their industrial processing. It has four groups, from the raw product (group 1) to the ultra-processed (group 4). Our guideUnderstanding the NOVA classificationcovers the whole thing in detail.

In a 2019 reference article, Monteiro and colleagues describe ultra-processed foods as resulting from several industrial processes: fractionation of foods into substances, chemical modification of these substances, assembly, frequent use ofcosmetic additives(flavourings, colourings, emulsifiers, sweeteners) and sophisticated packaging. The central idea: they are formulations made mainly from food derivatives, with little or no whole food.

The authors also stress that these products are designed to be profitable, ready to eat and highly appealing. It is a definition based onprocessesandingredients, not on nutritional composition.

How to recognise them on the label

The practical test proposed by the authors of NOVA is simple: look at the ingredient list. A product is classed as group 4 as soon as it contains at least one ingredient not found in a domestic kitchen, or an additive intended to make the product more appealing. The most telling signs:

Glucose-fructose syrup, maltodextrin, dextrose or other industrial sugars.
Isolated or hydrolysed proteins(from milk, soy, wheat).
Hydrogenated oilsor interesterified oils.
Flavourings, including "natural flavouring", which make a product more flavourful without any corresponding real ingredient.
Emulsifiers, thickeners, gelling agents(lecithins, mono- and diglycerides, gums, carrageenans).
Colouringsandsweeteners.
Along list, with technical names or E-numbers, rather than a few recognisable ingredients.

A single one of these markers is enough in theory. On additives, our articleAdditives: are they all a real danger?helps sort out those that are harmless from those that deserve more attention.

What studies show: associations, not proof of cause

In France, theNutriNet-Santécohort followed 44,551 adults aged 45 and over for a median of 7.1 years. In the analysis published in 2019 in JAMA Internal Medicine, each 10-point increase in the share of ultra-processed foods in the diet was associated with a 14% higher risk of death from all causes (risk ratio 1.14). The authors specify that the study is observational and its results need to be confirmed.

In 2024, anumbrella reviewpublished in the BMJ by Lane and colleagues pooled 45 meta-analyses. Adverse associations predominate: 32 of the health parameters studied (71%). According to the authors' grid, the evidence is rated "convincing" for cardiovascular disease mortality, type 2 diabetes and certain common anxiety or mental disorders, and "highly suggestive" for all-cause mortality, obesity or depressive symptoms. But GRADE quality is low or very low for the great majority of analyses (22 low, 19 very low, 4 moderate), and all the included studies are observational.

On the experimental side, the trial byKevin Hall(NIH, 2019) followed 20 adults over two two-week periods, one on an ultra-processed diet and the other on an unprocessed diet, with meals matched for calories, sugars, sodium, fibre and macronutrients. On the ultra-processed diet, participants consumed around508 kcal more per dayand gained around 0.9 kg, compared with a loss of around 0.9 kg on the other. The trial is controlled, but very short and small.

In November 2025, aseries of three Lancet articles, led by Monteiro and funded by Bloomberg Philanthropies, argued that an ultra-processed diet drives chronic disease. Several authors are behind the NOVA classification.

ANSES's position: caution

In an opinion published in January 2025,ANSESreviewed the literature. It concludes, with alow weight of evidence, that higher consumption of foods classed as ultra-processed under NOVA is associated with a higher risk of mortality and chronic disease (type 2 diabetes, overweight, obesity, cardiovascular disease, breast and colourectal cancers).

But the agency considers that there is no consensual definition, judges NOVA partly subjective because it relies on the presence of additives "without distinction", and considers that current classifications cannot be translated into health risks nor serve as a nutrition education tool. It recommends studying possible mechanisms: excessive food intake linked to the formulation, and the formation of neoformed substances.

The limits of the classification

Criticisms concern NOVA's precision. A team of researchers from the NutriNet-Santé cohort acknowledges that the classification has been criticised for its lack of robustness, and that assigning a group often relies on analysing the ingredient list, with possible inaccuracies.

The same work notes that some NOVA 4 products have a favourable nutritional profile (unsweetened flavoured water, certain canned pulses), while unprocessed foods such as homemade cake or pure fruit juice have a rather poor profile. Among NOVA 4 products rated A or B under the former Nutri-Score, frequent examples wereyoghurts, dairy drinks,sliced breadand rusks. The Nutri-Score update reduced the share of NOVA 4 bread rated A or B by around 70%.

Practical conclusion: NOVA and the Nutri-Score measure two different dimensions,processingon one side,nutritional qualityon the other. Looking at them together is more telling than relying on just one.

Everyday families concerned and simpler alternatives

Breakfast cereals and biscuits: replace with oat flakes or plain wholemeal bread, with a little fruit.
Flavoured yoghurts and dairy desserts: prefer plain yoghurt, flavoured with fruit or a little honey.
Sweetened or diet drinks: water, optionally infused, or freshly pressed fruit juice in moderate amounts.
Ready meals, nuggets, industrial pizzas: simple homemade versions, or plain frozen produce (vegetables, fish) to save time.
Cold cuts and sausages: read the list, knowing that additives are covered in our articleNitrites in cold cuts.
Bars, sauces and savoury biscuits: watch for palm oil, seePalm oil: real dangers?.

Many of these products also contain sugar and salt in quantities you would not suspect: seeHidden sugar and salt: where are they?. And to build simpler meals, our guideDigestion and immunity: friendly foodsoffers ideas.

Reduce, without guilt

The studies cited concern thesharethey occupy in the diet. The reasonable goal is agradual reduction, starting with products eaten every day, rather than a pursuit of perfection.

In Yumiq, the product page shows the number of additives detected and classes them as harmless, to watch or risky. The Yumiq-Score penalises additives flagged by EFSA (risk of overexposure) or ANSES in the Open Food Facts taxonomy, and the Yumiq star requires the absence of a risky additive. The NOVA group is one of the criteria taken into account, alongside the Nutri-Score, so that a product is not judged on a single indicator.