A packet of sweetened cereal can look respectable on a nutrition panel: fortified, low-fat, a few grams of fibre, maybe a vitamin claim. NOVA ignores the halo and reads the manufacturing history. If the food is built from refined fractions, cosmetic additives, flavour systems, emulsifiers, and industrial textures, it moves into Group 4. That missing label layer has predicted disease risk better than many shoppers expect.
Carlos Monteiro, a Brazilian public-health nutrition researcher, made the provocation in Public Health Nutrition in 2009: the problem was not only foods or nutrients, but processing. The system that grew from that idea, NOVA, sorts foods into four groups by the extent and purpose of industrial processing, from minimally processed foods to ultra-processed formulations.1
Group 1 is food close to its original form: fruit, vegetables, pulses, grains, milk, eggs, fish, plain meat, nuts, and seeds. Group 2 is what cooks use to prepare Group 1: oil, butter, sugar, salt, starch, and flour. Group 3 combines them in recognizable ways: cheese, pickles, canned beans, salted nuts, and simple breads. Group 4 is different. It is usually a formulation made from substances extracted or derived from foods, with additives for colour, flavour, mouthfeel, sweetness, durability, or convenience.2
The strongest causal clue comes from Kevin Hall's 2019 randomized controlled trial in Cell Metabolism. Twenty adults lived at the NIH Clinical Center and ate ultra-processed or unprocessed diets for two weeks, then switched. The meals were matched for presented calories, sugar, fat, fibre, sodium, and macronutrients. Participants could eat as much or as little as they wanted. On the ultra-processed diet, they consumed 508 extra calories per day and gained about 0.9 kg. On the unprocessed diet, they lost about 0.9 kg.3
Observational evidence is broader, though it cannot prove causation alone. A 2018 BMJ NutriNet-Sante cohort study of 104,980 French adults found that every 10% increase in ultra-processed food share was associated with higher overall cancer risk. A 2019 BMJ NutriNet-Sante cardiovascular paper followed 105,159 adults for a median 5.2 years and found higher cardiovascular, coronary heart disease, and cerebrovascular disease risk with higher UPF share. The common shorthand says "10 years"; the actual median follow-up in the cardiovascular paper was 5.2 years.4
The evidence base is now much larger. A 2024 BMJ umbrella review pooled 45 analyses from 14 meta-analyses, covering nearly 9.9 million participants. Greater ultra-processed food exposure was consistently associated with higher risk across mortality, cancer, mental, respiratory, cardiovascular, gastrointestinal, and metabolic outcomes. The review found no randomized trials to pool, which is why Hall's small inpatient trial matters so much.5
The 30-second test: if the ingredient list contains flavour systems, emulsifiers, thickeners, artificial sweeteners, colours, maltodextrin, protein isolates, modified starches, or ingredients you would not cook with at home, treat it as likely NOVA Group 4.
India's angle is not imported panic. It is domestic policy catching up to a changing food environment. A WHO India report released in 2023 analysed the growth of ultra-processed food sales and policy gaps. A 2024 BMC Public Health mapping study noted that Indian UPF sales rose from USD 0.9 billion in 2006 to USD 37.9 billion in 2019, across products such as sugary drinks, chips, biscuits, bread, and ready-to-eat or ready-to-cook meals.6
India also has its own processing classification in the 2024 ICMR-NIN Dietary Guidelines for Indians. NIN describes primary, secondary, and tertiary processing, defines ultra-processed foods as products with extensive industrial processing and additives, and uses Category A, B, and C, where Category C includes UPFs. But this is dietary guidance. FSSAI's front-of-pack path has focused on Indian Nutrition Rating and high fat, sugar, and salt disclosure. As of May 2026, Indian packaged foods still do not have to print "NOVA Group 4" or "ultra-processed" on the front label.7
NOVA is useful, not perfect. It can group very different products together: a fortified wholegrain breakfast cereal, a cola, a vegan meat alternative, and a packaged biscuit can all land in Group 4. That makes some nutrition scientists uneasy, and fairly so. Nutrient quality, affordability, access, and cultural context still matter.
But the pushback has its own label problem. A 2018 World Nutrition paper mapped 32 materials criticising NOVA and identified 38 authors; 33 had relationships with the ultra-processed food and beverage industry. That does not automatically make every criticism false. It does mean readers should ask the same question they ask of the food itself: who manufactured this argument?8
Pick one packaged food you buy every week and classify it before reading the nutrition panel. First, ask if it is a recognizable food, a cooking ingredient, a simple processed food, or an industrial formulation. Then scan for cosmetic additives: flavours, colours, emulsifiers, thickeners, sweeteners, isolates, modified starches, maltodextrin, or ingredient names that would never appear in your kitchen. If it is Group 4, you do not have to ban it. Move it from daily default to occasional convenience.