Nutri-Score or NOVA: which should you look at?

1. The fundamental difference is the question each system asks

Nutri-Score is a voluntary front-of-package label with five grades, from A to E, calculated per 100 g or 100 ml. The current French algorithm balances components to limit against favorable components, with specific rules for several food categories. It primarily answers one question: among products that can realistically replace one another, which has the more favorable nutritional profile according to this model? Public Health France — official overview of Nutri-Score, Légifrance — current French algorithm

NOVA places foods in four groups according to the nature, extent, and especially the purpose of their processing. Its authors specify that “NOVA” is a name, not an acronym. The groups do not form a continuous score: group 2 includes processed culinary ingredients such as oil, sugar, and salt, and is not simply one step between “good” and “bad.” Monteiro et al., 2016 — formalization of NOVA’s four groups, Monteiro et al., 2019 — identification criteria

Both systems aim to simplify information, but they summarize neither the same inputs nor the same unit of analysis. Treating them like two thermometers measuring the same temperature creates most of the apparent contradictions.

• Understand the definition and history of Nutri-Score →
• Discover the origin and four groups of the NOVA classification →

2. Comparison table: two tools with different blind spots

This table does not declare one system superior. It matches each tool with the decision it can genuinely inform, then shows what that tool leaves outside its scope.

3. How can a product with Nutri-Score A or B be NOVA 4?

A manufacturer can formulate a product that is low in salt or sugar and high in fiber or protein, earning a favorable Nutri-Score profile, while also using food fractions, flavorings, or additives characteristic of a NOVA 4 formulation. Both results can follow their respective rules: one describes selected nutrients, while the other describes processing.

The reverse also occurs. Oil, cheese, or a traditionally processed food may receive a less favorable grade per 100 g without belonging to group 4. Its energy density, salt, or saturated fat does not disappear because the ingredient list is short; its place in a diet also depends on the amount and frequency consumed.

The mismatch is therefore not proof of manipulation. It is additional information. Make it useful by asking: is this an alternative to another product with the same role, what does the ingredient list show, how much will I actually eat, and what will this choice replace?

• Calculate the current Nutri-Score step by step →
• Identify ultra-processed foods without a simplistic rule →

4. The evidence suggests complementarity, not one combined truth

The Italian Moli-sani cohort followed 22,895 adults and recorded 2,205 deaths. A less favorable nutritional profile and greater consumption of ultra-processed foods were each associated with higher mortality. Adjusting for both dimensions weakened the nutritional-profile association more than the ultra-processed-food association. The result suggests partial overlap without implying that the two dimensions are identical. Bonaccio et al., 2022 — Moli-sani cohort

The study remains observational. It does not prove that a label causes or prevents death, or that NOVA alone explains a health difference. Habits, illnesses, income, and measurement error may also contribute to the observed association.

An online experiment involving 21,159 people found that adding an ultra-processed-food banner to Nutri-Score improved participants’ simultaneous understanding of nutritional profile and processing level. Participants were already knowledgeable about nutrition, the experiment mainly measured understanding and intentions, and it tested neither lasting behavior nor clinical outcomes. Srour et al., 2023 — online testing of NS2.0 format

The proportionate conclusion is modest: presenting two dimensions can reduce immediate confusion. Combining them does not create a medical algorithm or causal evidence about health.

5. Real controversies are not evidence of fraud

Nutri-Score is criticized for its 100 g basis, conventions, exceptions, influence from economic stakeholders, and grade changes after revisions. NOVA is debated because of sometimes subjective boundaries, the diversity within group 4, and the difficulty of separating processing from composition, texture, food matrix, and social context. These debates are documented and legitimate.

ANSES concluded in 2025 that there was no consensus definition of ultra-processed foods and that applying existing definitions could be subjective. It rated the overall weight of evidence for several health relationships as low. “Low” does not mean “no association”; it describes confidence in causal interpretation and mechanisms. ANSES, 2025 — characterization and impacts of ultra-processed foods

On June 16, 2026, France’s Council of State rejected several arguments raised by Lactalis against the Nutri-Score order, but referred two questions of European law to the Court of Justice of the European Union and stayed its decision. As of July 26, 2026, the CJEU had not ruled, and the French order remained in force. It would be inaccurate to say that Lactalis won or that Nutri-Score was declared illegal. French Council of State — Groupe Lactalis decision No. 508230

Conflicts of interest, lobbying, and involvement by a system’s promoters should be disclosed and considered alongside study protocols. They do not automatically prove data fabrication. Our review identified no formal finding of fraud, foundational retraction, or demonstrated corruption of either system; the accurate term is “documented controversies.”

• Review the evidence, limitations, and controversies surrounding Nutri-Score →
• Examine NOVA’s reliability and reproducibility →

6. A two-part method for reading products in the grocery aisle

Start by defining the product’s role: compare two breads, cereals, sauces, or prepared meals that could genuinely replace one another. Without a plausible substitution, comparing grades loses much of its value.

First, use Nutri-Score to spot a difference in nutritional profile, then read the nutrition facts to see whether it comes from salt, sugars, saturated fat, fiber, or energy. Second, examine the ingredient list and formulation. Look for the original food, culinary ingredients, food fractions, and additives—without treating every additive as proof of harm.

Finish with what neither system knows: your serving, frequency, budget, allergies, enjoyment, food waste, access to a kitchen, and the rest of your meals. A realistic decision you can repeat is better than a perfect rule you cannot sustain.

• Define the product being replaced and compare foods with the same use.
• Read the Nutri-Score as a nutritional filter, not a license.
• Read NOVA as a description of formulation, not a toxicity score.
• Check the nutrition facts, ingredient list, and serving actually consumed.
• Consider the substitution and your diet across the week.

7. Three examples that avoid automatic verdicts

For bread, begin by comparing two products intended for the same meal. The grade may highlight differences in fiber or salt; NOVA may differ according to the recipe and additives. Words such as “whole grain” or “artisan” are not enough to assign either grade without precise data.

For yogurt or dessert, distinguish the base food, sugar content, and formulation. A simple appearance does not guarantee NOVA 1, and a favorable Nutri-Score does not exclude NOVA 4. Consider what the product replaces: a sweeter dessert, plain yogurt, a meal, or nothing at all.

For a protein bar or drink, protein content does not erase energy, sugars, sweeteners, salt, or formulation. Conversely, a NOVA 4 classification cannot tell you whether one serving meets a practical need. Never guess either result from marketing or product category alone.

8. For athletes, neither system calculates individual needs

Nutri-Score does not know your energy expenditure, protein target, workout timing, or exercise duration. NOVA does not measure the grams of protein, carbohydrate, sodium, or energy in a serving. You therefore cannot build a sports-nutrition strategy by collecting A grades or automatically excluding every group 4 product.

A convenient product may serve a specific purpose around training, while a minimally processed food may be unsuitable just before exercise because of its volume or fiber. That situational value does not make the product an ideal foundation for the entire diet.

Use both systems to inform comparisons, then calculate intake using actual servings and assess tolerance. A medical condition, allergy, pregnancy, eating disorder, or demanding athletic goal requires an individualized decision that goes beyond these classifications.

9. For health claims, separate associations, trials, and mechanisms

The 2024 BMJ umbrella review on ultra-processed foods brought together 45 analyses and nearly 9.9 million participants. Direct associations appeared for 32 outcomes, but only four analyses reached moderate GRADE certainty; 22 were rated low certainty and 19 very low certainty. An association described statistically as “convincing” is not the same as causal evidence. Lane et al., 2024 — umbrella review of ultra-processed foods

Hall’s metabolic-ward trial found that 20 adults consumed approximately 508 kcal more per day and gained 0.9 kg during a two-week ultra-processed diet, compared with a loss of 0.9 kg during a two-week unprocessed diet. It demonstrates a short-term effect of those menus without isolating one intrinsic feature of processing or measuring chronic disease. Hall et al., 2019 — randomized metabolic-ward trial

For Nutri-Score, the evidence strongly supports immediate comprehension, while effects on actual purchases are more modest. Cohort studies support the relevance of the underlying nutritional profile, not a causal effect of the label itself. For NOVA, trials and cohorts support a serious health question, but product subgroups, mechanisms, and classification errors limit universal verdicts.

• Read what research says about ultra-processed foods and health →

10. The useful conclusion: compare, complete, and contextualize

Nutri-Score is generally the first filter when you are choosing between substitutable products. NOVA adds a formulation question that the grade does not ask. If both point in the same direction, the decision may be simple; if they diverge, that is not a failure but a prompt to check the ingredients, serving, and substitution.

Do not expect one score or group to summarize the quality of your week. Raw or minimally processed foods, home cooking, variety, and products suited to real constraints can form the foundation; evaluate processed foods according to their role, frequency, and what they replace.

Finally, the rules evolve. The new Nutri-Score algorithm applied in France since March 2025 changed some grades, while the WHO is still preparing a global recommendation on ultra-processed foods. Always date legal and institutional statements before turning them into advice.

• See why some grades changed under the new Nutri-Score →