
By Yumiq
Which nutritional score is the most reliable?
Nutri-Score, NOVA, Green-Score: none are perfect. What are their blind spots, and how can they be combined to evaluate a product?
Why the question is wrongly framed
Nutri-Score, NOVA, Green-Score: we want to know which one to trust. Yet these three benchmarks do not answer the same question. The first evaluates thenutritional profile, the second thedegree of processing, the third theenvironmental impact. Asking which is the most reliable is like asking whether a thermometer is more reliable than a scale.
The real question is therefore: for which purpose is each score robust, and where do its proofs end? This article separates what is established from what is debated. For a side-by-side comparison, seeAll scores compared: Nutri-Score, NOVA, Green-Score.
Nutri-Score: the most validated, but not without limits
This is the score that has been studied the most. It is based on the nutrition table and on a few elements from the ingredient list, according toSanté publique France. What is established:
- A link to health at the population level.In the European EPIC cohort (501,594 participants, 10 countries, followed from 1992 to 2015), people who consumed more lower-scored foods had higher mortality, from all causes combined and for several causes (cancers, circulatory, respiratory, digestive diseases), after adjustment for many factors. These are associations, not proof of causality (Inserm, BMJ 2020).
- A link to cardiovascular risk for the new version.A September 2024 Inserm study evaluated the new algorithm and reported that consumption of lower-scored foods is associated with more cardiovascular diseases (according to the Santé publique France summary).
- An effect on products.According to the same summary, a 2022 US study observed improved nutritional quality of new products after the introduction of the logo, and an OECD modelling exercise ranks Nutri-Score first among several logos compared.
These results largely come from teams closely involved in the design of the score, particularly around the NutriNet-Santé study. They are published and peer-reviewed, but a literature review (van der Bend et al., 2022) highlights limitations: the score effectively classifies foods by families without demonstrating that products from different classes actually differ in nutritional quality; there is no gold standard to validate it, studies do not yet allow definitive predictions on disease risk, and its concordance with each country's dietary guidelines remains to be assessed. Other work, notably that of Julia and Egnell on the NutriNet-Santé cohort, points in the same general direction, but we do not detail it here.
The algorithm has changed: proof that it is being corrected
An independent scientific committee, convened for the first time in February 2021, proposed changes. Its reports date from June 2022 (solid foods) and February 2023 (drinks), validated by the transnational steering committee. The new rules have applied in Germany, Belgium, Switzerland and the Netherlands since 1 January 2024, in Luxembourg since March 2024, and in France since 16 March 2025, with a two-year period to adapt packaging.
What changes, according to Santé publique France:
- Oils low in saturated fats (olive, rapeseed, walnut) move from C to B;
- Oily fish receive higher ratings, and wholegrain starchy foods are better distinguished from refined ones;
- Salty and sugary products are rated more strictly;
- Water is the only beverage rated A, and drinks containing sweeteners move from B to C-E;
- Milk and plant-based drinks follow the beverage algorithm.
Between 30 and 40% of products change their rating. These adjustments respond to valid criticisms (olive oil, sweetened drinks): a score that changes after review is more a positive sign than a flaw. However, consumer press outlets argue that sweeteners are penalised only in beverages, whichTest Achatsdescribes as a "missed opportunity".
What remains debated regarding Nutri-Score
- The degree of processing is not included in the calculation.A highly processed product can achieve a good score if it is low in sugar, salt and saturated fats. This is the purpose of NOVA.
- Its application to traditional products.Italy contests the logo in the name of its traditional products and advocates for another system, Nutrinform, according to press reports. This disagreement is as much political as it is scientific.
- Its voluntary nature.The logo is applied by producers on a voluntary basis, Santé publique France reminds us: the absence of a logo says nothing about the product.
At European level, the Commission announced in its "Farm to Fork" strategy a review of consumer information rules, with mandatory and harmonised front-of-pack nutrition labelling. Its official page still describes this project as a proposal under consideration, without a timetable. Media outlets reported in early 2025 that the Commission would abandon imposing an existing system: a press report not confirmed by an official document. Therefore, to date, there is no mandatory European logo.
NOVA: a useful framework, but a debated classification
NOVA is the most widely used framework in studies on ultra-processed foods. In its January 2025 opinion, theANSESnotes that there is no consensus definition of these foods, and that NOVA, by counting additives and substances without distinguishing between them, has a partly subjective application.
Its systematic review concludes that there is an association, with a low level of evidence, between high consumption of ultra-processed foods and mortality or chronic diseases (type 2 diabetes, obesity, certain cancers). For the agency, these classifications cannot, as they stand, be translated into health risks. Established: NOVA accurately describes a degree of processing. Debated: what this degree implies for health, and the exact boundary between the groups.
The Green-Score: an experimental indicator
The Green-Score, from A+ to F, summarises a product's environmental impact over its life cycle (greenhouse gases, water, eutrophication, land use, biodiversity). Its official documentation presents it as anexperimental indicator, as part of a pilot scheme run by the Ministry of Ecological Transition, and as a decision-making aid, not as a definitive measure. It is useful for comparing similar products, but it is newer and less validated than Nutri-Score, and it says nothing about health.
No score is perfect: how to combine them
Each score has a blind spot that the others fill. A combined approach relies on four simple rules:
- Nutri-Scoreto compare products within the same family on a nutritional basis (two breakfast cereals, two ready meals);
- NOVAto identify highly processed products, even those rated highly by Nutri-Score;
- Green-Scoreto differentiate between products of comparable nutritional quality, bearing in mind its experimental status;
- the ingredient listto make a final judgment: short, recognisable, with few additives.
An A Nutri-Score with a NOVA 4 and many additives warrants a second look; a D on an indulgent product consumed rarely does not raise the same concerns as a D on a daily staple. No score replaces frequency of consumption or the context of the meal.
Why Yumiq combines them
Since no single score answers everything, theYumiq-Scorebrings together several dimensions: Nutri-Score, NOVA group, Green-Score, a penalty for additives flagged by EFSA (risk of overexposure) or ANSES in the Open Food Facts taxonomy, a bonus for verifiable quality labels, and an average of community reviews. Each dimension remains available for detailed consultation on the product page. For details on how the first of these indicators works, seeUnderstanding Nutri-Score: calculation, grades A to E and limitations.
This combination does not erase the above debates: it avoids relying on a single perspective and a single weakness.
Sources
- Santé publique France - Nutri-Score
- Santé publique France - Signing of the Nutri-Score decree (17 March 2025)
- Inserm - Consumption of foods with lower Nutri-Score ratings and mortality (BMJ, 2020)
- van der Bend et al. - The Nutri-Score algorithm: Evaluation of its validation process (Frontiers in Nutrition, 2022)
- European Commission - Food information to consumers legislation
- ANSES - So-called ultra-processed foods (2025)
- Environmental score - Green-Score documentation
- Test Achats - Nutri-Score: the changes
Frequently asked questions
- Which nutritional score is the most reliable?
- None are reliable for everything, as they measure different things. Nutri-Score is the most validated for nutritional profile, NOVA describes the degree of processing (a classification questioned by ANSES), and Green-Score is an experimental environmental indicator. They should be read together.
- Is Nutri-Score scientifically validated?
- It has been the subject of numerous studies. For example, the European EPIC cohort (501,594 participants) links higher consumption of lower-scored foods to higher mortality. These are associations, not proofs of causality, and a 2022 review also notes validation limitations.
- Why have olive oil and sweetened drinks changed their scores?
- The algorithm was revised by a scientific committee. Since March 2025 in France, oils low in saturated fats such as olive oil move from C to B, and drinks containing sweeteners move from B to a score between C and E. Water is the only drink rated A.
- Is there a mandatory European score?
- No. Nutri-Score is applied on a voluntary basis. The European Commission has announced harmonised front-of-pack nutrition labelling, but its official page still presents it as a proposal under consideration, with no timetable.