Few food ideas have spread as fast, or as loosely, as the claim that ultra-processed food is harmful in itself. The phrase now carries the weight of a verdict. Yet the strongest experiments behind it are narrower than the slogan, and they do not all point in the same direction. Read together, they describe a real effect on body weight while leaving the mechanism genuinely open. That distinction matters, because the answer to "is it the processing, or what the processing does to the food's calories?" changes what a person, or a government, should actually do.

What the controlled trials measured

The cleanest evidence comes from studies in which researchers supply every meal. In the best known, Kevin Hall and colleagues at the U.S. National Institutes of Health confined 20 adults to a metabolic ward for a month and fed them, in random order, two weeks of ultra-processed food and two weeks of minimally processed food. The menus were matched for presented calories, sugar, fat, fiber, and macronutrients, and participants were told to eat as much or as little as they wanted. On the ultra-processed diet they consumed about 508 more kilocalories a day and gained roughly 0.9 kilograms; on the unprocessed diet they lost about the same amount.

A larger and longer trial, the UPDATE study led by Samuel Dicken at University College London and published in Nature Medicine in August 2025, moved the question out of the ward and into ordinary life. Fifty-five adults in England who habitually got at least half their calories from ultra-processed food were given two eight-week diets, in random order, both built to follow the UK Eatwell Guide: one of minimally processed food, one of ultra-processed food, with all meals delivered to their homes. On both diets people lost weight. The difference was that they lost more on the minimally processed diet — a within-person difference of about one percentage point of body weight over eight weeks, a modest but statistically significant gap. The authors estimated that, scaled over a year, the gap would amount to a meaningfully larger weight reduction on the less-processed diet.

Two things make these results worth taking seriously. They are randomized and controlled, which the much larger body of observational studies linking ultra-processed food to disease is not. And the UPDATE trial was designed to isolate the degree of processing: by holding the dietary guidelines constant, it aimed to compare foods that differed mainly in how processed they were, rather than in the quality of the overall diet. But the two diets were not matched for energy density or nutrient profile — a gap that critics would later seize on, and that the trial's own authors acknowledged.

The complication the meta-analysis raises

The trials show a difference in weight. They are less conclusive about why. A June 2026 systematic review and meta-analysis by Eric Robinson and colleagues, posted as a preprint on medRxiv and not yet peer-reviewed, pooled ten randomized trials and found a consistent signal for weight gain on ultra-processed diets, but a muddier one for calorie intake. (For transparency: Robinson is also a co-author of one of the published critiques of the UPDATE trial discussed below.) The decisive detail is a subgroup split. In trials where the ultra-processed foods were more energy-dense than the comparison foods, intake rose clearly. In the few trials where the foods were matched for energy density, the difference in calorie intake essentially vanished.

That pattern points toward a mechanism that is less mysterious than "processing is toxic." Ultra-processed foods tend to be soft, calorie-dense, and easy to eat quickly, and the meta-analysis found that eating rate was the factor most strongly tied to higher intake. In Hall's ward study, participants ate faster on the ultra-processed diet, and protein intake stayed flat while carbohydrate and fat intake climbed — a pattern consistent with the idea that people keep eating to reach a protein target. None of this requires an additive or an industrial ingredient to be doing hidden harm. It is enough that the foods are engineered to be consumed in large amounts with little effort.

The practical consequence is uncomfortable for a simple boycott. If the harm runs mainly through calorie density and eating speed, then a diet of carefully chosen, lower-density ultra-processed products could behave much like a minimally processed one — and indeed the UPDATE participants lost weight on their ultra-processed diet when it was designed to meet dietary guidelines. If, instead, some feature of processing acts independently of calories, then no amount of label-reading fully solves the problem. The current evidence cannot settle which is true.

What remains unresolved

Several limits keep the question open. The trials are small and short; the longest runs to a few months, and weight effects over years are inferred, not measured. The meta-analysis is a preprint, so its pooling choices have not yet survived peer review, and it notes that the trials did not match diets for nutrient profile. That mismatch is now the subject of open scientific dispute: in January 2026, Nature Medicine published three Matters Arising challenging the UPDATE trial — from David Ludwig and colleagues, from Robinson and Ciaran Forde, and from Wang and Peng — arguing that the two diets were nutritionally mismatched, with the ultra-processed diet nominally lower in fat and energy density, in ways that could explain the weight difference without invoking processing at all. The trial's authors replied that they had deliberately not equated energy density, since real-world ultra-processed diets differ on exactly that dimension. The disagreement is not settled. The NOVA classification that defines "ultra-processed" groups together foods as different as mass-produced bread and sugary drinks, which makes a single mechanism hard to pin down. And almost all of the controlled feeding work has been done in wealthy countries, on relatively small samples, leaving open whether the findings hold across diets and populations.

What the evidence does support is narrower than the public claim but still useful: when people are given ultra-processed food and minimally processed food that follow the same healthy guidelines, they tend to eat more and weigh more on the former. The stronger claim — that processing itself, independent of calories and nutrients, is the cause — is plausible but not yet proven. For a reader, the durable takeaway is not to fear a category, but to notice the two things the trials actually moved: how calorie-dense the food is, and how fast it is eaten. Those are levers a person can act on regardless of where the final science lands.

Sources