Nutri-Score 2023 tested on 408,399 EU food labels

On September 15, 2026, BMJ Public Health published a cross-sectional study that applied the updated 2023 Nutri-Score front-of-pack algorithm to 408,399 branded foods from 14 European countries. The authors report that the 2023 version graded foods consistently across countries and in line with food-based dietary guidelines, and that it moved the most common grade down for categories such as high-sugar breakfast cereals, pizza, pastries and reduced-sugar colas with sweeteners. The grade is built from nutrients per 100 g and fruit, vegetable and legume content only, so it says nothing about additives or processing.

Key takeaways

• On September 15, 2026, researchers on the Nutri-Score Scientific Committee published an assessment of the 2023 algorithm applied to 408,399 branded products from 14 EU countries.
• Compared with the 2015 version, the most common grade moved down for high-sugar breakfast cereals, pizza, ready-to-eat meals, pastries, yoghurt and reduced-sugar colas with sweeteners, while olive and canola oils reach B and nuts and seeds shift toward A.
• Nutri-Score uses nutrients per 100 g plus fruit, vegetable and legume content; it does not account for additives or processing level.
• The authors flag remaining gaps, such as dried pasta, including refined pasta, being consistently graded favourably. Nutri-Score is voluntary.

What happened

On September 15, 2026, BMJ Public Health published a multi-country cross-sectional analysis titled "Does the Nutri-Score 2023 algorithm provide consistent classification across European food supplies? A cross-sectional study using branded food composition data from 14 countries" (Julia C, Vasseur T, Alexiou H, et al.; doi: 10.1136/bmjph-2026-005015; full text at PMC13583817). The authors are members of the Nutri-Score Scientific Committee, the body that designed the 2023 update, so this is not an independent evaluation.

The research evaluated branded food composition records from 14 European countries: Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Poland, Spain, and Sweden. The data were 2022-2023 records from Euromonitor, collected by web scraping a varied selection of retailer websites. After excluding categories outside the scope of Nutri-Score and products with missing or outlier nutrition data, 408,399 products remained.

This investigation was strictly an evaluation of label classification consistency, not a health-outcome study. As confirmed by the ethics statement, the dataset comprised exclusively food composition data. No human subjects, dietary intakes, or personal health records were evaluated.

What changed

The authors applied both the original 2015 Nutri-Score criteria and the revised 2023 algorithm to the full dataset. Nutri-Score sorts products into five grades, from A (most favourable) to E (least favourable), using points for unfavourable and favourable components per 100 g. The inputs used in the study were energy, saturated fat, sugars and salt on one side, and protein, fibre and fruit, vegetable and legume content on the other.

Comparing the two algorithms on the same products, the authors report these shifts in each category's most common (modal) grade and distribution:

• High-sugar breakfast cereals shifted their modal grade from C down to D.
• Pizza moved from C to D.
• Ready-to-eat meals shifted from a modal grade of B down to C.
• Pastries moved from D to E.
• Yoghurt moved from B to C.
• Reduced-sugar colas made with non-nutritional sweeteners shifted from B down to C.
• Olive and canola oils reach grade B.
• Nuts and seeds shift toward grade A.

Across the 408,399 branded products, the authors reported that greater than 90% of frozen fruits and vegetables and greater than 95% of pulses fell into favorable grades A and B. Conversely, more than 90% of discretionary categories, such as chocolate confectionery and sweet biscuits, landed in unfavorable grades D and E.

Who is affected

The findings cover packaged foods sold across the 14 analyzed European markets. Sample sizes per country ranged from 6,949 products in Greece to 69,651 in France, and exclusion rates ranged from 16% in Greece to 45.4% in Denmark.

The most heavily represented supermarket categories in the final analyzed dataset were chocolate and chocolate confectionery (34,819 products), sweet biscuits (25,917 products), and savoury snacks (21,175 products). The distribution across other major staples showed clear boundaries: refined bread scored mostly C or D (93.8% of items), regular-fat unprocessed cheese scored mostly D (79.6%), whole milk scored mostly C (93.1%), skimmed or semi-skimmed milk landed in grade B (>99%), colas were 83.62% grade E, and processed red meat products reached 54.11% grade E.

The authors say cross-country differences "reflected underlying variability in nutrient composition of products", giving the example that salt content in refined breads ranged from 1.05 to 1.23 g/100 g across countries.

What it means for shoppers

A front-of-pack grade is a quick shortcut, not a full read of the package. Nutri-Score summarises a handful of nutrients per 100 g; it does not look at how a product is made. A shopper reading about Seed Oils on a Label: What It Tells You, What It Doesn't will see the limit: olive and canola oils reach B under the 2023 update, but because the grade is built from nutrient values, it does not tell you how an oil was processed.

Nutri-Score does not account for additives such as emulsifiers, stabilisers or preservatives. A product with several of them can still grade well if its nutrient profile is favourable. Shoppers reviewing Emulsifiers on Ingredient Lists Explained: Lecithin to Polysorbate 80 or examining E Numbers Explained: The EU Code for US Additives will find that none of these ingredients count toward the letter grade.

The authors also name gaps in the algorithm itself. They write that "the consistently favourable classification of dried pasta, including refined varieties, reflects a known limitation of the algorithm", and list whole-grain pasta and rice among the product groups where "residual challenges" remain.

The background

According to the paper, Nutri-Score has been adopted as the official front-of-pack scheme by Belgium, France, Germany, Luxembourg, the Netherlands and Switzerland, with use announced in Spain. It is voluntary, not mandatory, and it is not used in the US.

The study was funded by the members of the Steering Committee of the Nutri-Score, which paid for the data acquisition and analysis. The paper states that "the funders had no role in the design, data collection, data analysis and reporting of this study" and declares no competing interests. Readers should still weigh that the authors helped design the algorithm they are evaluating.

In the United States, the Food and Drug Administration (FDA) has pursued a separate path. Rather than a single summary letter grade, the FDA has proposed a front-of-package Nutrition Info box that describes saturated fat, sodium and added sugars as "Low," "Med," or "High" (FDA docket FDA-2024-N-2910, with the public comment period extended to July 15, 2025). The proposal has not been finalized, and unlike Nutri-Score it does not give one summary grade.

What to watch next

Nutri-Score remains voluntary in the countries that adopted it, and the paper itself argues the 2023 algorithm could work as an EU-wide label. Whether that happens is not something this study decides.

Shoppers must also keep the research boundaries in mind. The study authors noted that retailer web-scraping datasets cannot guarantee complete market representativeness for every nation, some product misclassifications cannot be excluded, and the underlying Euromonitor product records are proprietary rather than publicly accessible. Most importantly, the study did not measure whether consuming A-grade foods instead of D-grade foods changes long-term health outcomes.