Ultra-processed food dominates headlines but what does the science actually say? Here’s a clear-eyed look at the evidence, the nuance and the real risks.
In the summer of 2023, a study landed in a major nutrition journal and briefly upended breakfast tables across the English-speaking world. Ultra-processed foods, it concluded the chicken nuggets, the sliced white bread, the cereal bars, the flavoured yoghurts were associated with a range of serious health outcomes, from cardiovascular disease to depression. The headlines were immediate and unequivocal. Then came the counter-argument. Scientists, food industry representatives and not a few frustrated nutritionists pointed out that the evidence was largely observational, that correlation was not causation, and that the category of “ultra-processed food” was so expansive as to be almost meaningless.
Both responses, in their own way, missed the point. The ultra-processed food debate is genuinely important, scientifically complex and frequently misrepresented in both directions. Here is what the evidence actually says.
The NOVA Classification: Understanding the Battleground
To understand the UPF debate, you first need to understand the classification system at its centre. The NOVA framework, developed by Brazilian epidemiologist Carlos Monteiro and his team, divides food into four groups based on the degree and nature of processing involved. At one end: unprocessed or minimally processed foods. At the other: ultra-processed products, defined as “industrial formulations made entirely or mostly from substances extracted from foods, together with additives and other substances.”
It is important to be clear about what NOVA is and what it is not. It is not a nutritional classification system. A food can be high in fibre, low in saturated fat and still be classified as ultra-processed under NOVA. This does not mean NOVA is wrong or useless. It means it is measuring something specific: the nature of industrial transformation and the social and behavioural context in which food is produced and consumed.
What the Studies Actually Show
The epidemiological evidence linking UPF consumption to negative health outcomes is substantial, and it is growing. The NutriNet-Santé cohort study in France, involving more than 100,000 participants, found associations between higher UPF consumption and increased risk of cancer, cardiovascular disease and all-cause mortality. A UK Biobank study of nearly 200,000 adults found a correlation between UPF intake and higher rates of anxiety and depression. These are not small studies. The consistency across populations and research groups is notable, and most epidemiologists take it seriously.
The problem is what these studies can and cannot tell us. All are observational they track what people eat and what happens to them over time, but they cannot prove that the food itself caused the outcome. People who eat more ultra-processed food also tend, on average, to have lower incomes, exercise less, sleep worse and experience higher levels of stress. The best researchers in the field go to considerable lengths to adjust for these confounders, but it is never possible to adjust for everything.
The Mechanistic Question
Where the science gets more interesting and more contested is in the question of mechanism. Why would ultra-processed food specifically harm health, beyond the fact that it tends to be energy-dense and nutrient-poor? Several hypotheses are under active investigation. One centres on food matrix disruption: the idea that processing breaks down the physical structure of food in ways that affect how the body absorbs nutrients and calibrates appetite. A whole almond and the same almond ground and extruded into an industrial snack product may have similar nutritional profiles on paper but behave differently in the body.
Another focuses on additives emulsifiers, artificial sweeteners, colourings and flavour enhancers which have been shown in animal and some human studies to affect the gut microbiome, potentially disrupting metabolic regulation and inflammatory pathways. A 2022 study in Cell found that certain common emulsifiers altered gut microbiota composition and increased intestinal permeability in human participants.
A third hypothesis is behavioural: that ultra-processed foods are engineered to override normal satiety signals, leading to overconsumption regardless of their specific ingredients. The combination of salt, fat and sugar in precise ratios is not accidental; it reflects decades of food science research aimed at maximising what the industry calls “palatability.”
Where the Consensus Is Heading
Few nutrition scientists would now argue that ultra-processed food is irrelevant to public health. The volume and consistency of the evidence is too substantial to dismiss. Where reasonable disagreement persists is in the mechanism, the magnitude of harm and the appropriate policy response.
What is increasingly clear is that the UPF category captures something real something about the relationship between industrial food production, appetite regulation and health even if it does not capture it perfectly. The nuance that is often missing from headlines is that not all ultra-processed foods are equal. A plant-based burger made from legumes occupies a different nutritional position from a potato chip designed to be impossible to stop eating.
The most useful takeaway for consumers is probably this: the evidence suggests that diets built substantially around whole and minimally processed foods are consistently associated with better health outcomes. That is considerably more robust than most dietary advice that preceded it.