Ultra-processed food and weight gain: the one controlled trial that proves causation

Ultra-processed food and weight gain: the one controlled trial that proves causation

Most of what you read about ultra-processed food and weight gain comes from studies that can only show a correlation: people who eat more packaged, ready-made food also tend to weigh more. That’s useful, but it can’t tell you whether the food is causing the weight gain, or whether something else about the people eating it explains the link. The ultra-processed food weight gain study that changed this conversation is different. In 2019, researchers at the US National Institutes of Health (NIH) ran an actual controlled experiment, and what they found gave nutrition science something it rarely has: a plausible causal answer, not just an association.

This article explains what that trial did, what it found, what a 2025 follow-up analysis adds, and – just as importantly – what it doesn’t settle. The exact mechanism behind the effect is still debated, and a separate, more recent trial has drawn its own scientific pushback. Both sides of that debate are worth understanding before you draw conclusions about your own diet.

What the Hall 2019 trial actually did

The study, led by NIH researcher Kevin Hall and published in Cell Metabolism, took 20 adults and housed them at the NIH Clinical Center for four weeks. This kind of “metabolic ward” study is expensive and logistically hard, which is exactly why so few of them exist – but it’s also why the results carry unusual weight.

Each participant ate two weeks of an ultra-processed diet and two weeks of an unprocessed, whole-food diet, in random order, with everyone crossing over to try both. Crucially, the two diets were matched for calories offered, energy density, sugar, sodium, fibre, and macronutrient split. Meals were designed so that, on paper, neither diet had an obvious nutritional advantage. Participants weren’t told to eat more or less – they had three meals a day plus snacks freely available and were simply told to eat as much or as little as they liked.

That “ad libitum” design is the trial’s real strength. Because the food environment was controlled and only the level of processing differed, any difference in what people actually ate couldn’t easily be blamed on differing nutrient content. It could plausibly be attributed to processing itself.

What did the ultra-processed food study find?

When given free access to an ultra-processed diet, participants ate around 508 kcal a day more than when given the unprocessed diet – roughly 20% more calories, driven mainly by extra carbohydrate (about 280 kcal/day) and fat (about 230 kcal/day), not protein. Over the two-week period, they gained an average of 0.9 kg on the ultra-processed diet and lost 0.9 kg on the unprocessed diet.

In short: same people, same amount of time, diets matched for calories and nutrients on paper – yet the ultra-processed version led to measurably more eating and more weight gain. That gap between “matched on paper” and “different in practice” is the whole story. Something about how the ultra-processed foods were formulated or presented was driving people to eat more without them noticing or intending to.

What the 2025 follow-up adds: energy density and eating rate

A 2025 mechanistic analysis in Cell Metabolism, revisiting the original trial data, dug into why the ultra-processed diet drove overeating. One striking figure: participants consumed the ultra-processed food at around 48 kcal per minute, versus 31 kcal per minute on the unprocessed diet – over 50% faster. That’s a meaningful clue, because eating rate and energy density (calories packed into a given weight or volume of food) are both independently linked to how much people eat before feeling full, regardless of how “hyperpalatable” a food tastes.

This matters because the simplest explanation – that ultra-processed foods are just engineered to taste irresistibly good – isn’t the whole picture. The 2025 analysis suggests that softer textures, easier chewing, and calorie-dense formulations may let people eat faster and get more calories in before their body’s fullness signals catch up, independent of how much they enjoy the taste. It’s a more precise, more mechanistic explanation than “hyperpalatability” alone, though the two ideas aren’t mutually exclusive.

What’s still debated

None of this means the mechanism is settled, and it’s worth being honest about that. Researchers are still working out how much of the effect comes from energy density, how much from eating rate, and how much from other factors like reduced chewing time or the sheer variety of flavours on offer.

The debate has also been sharpened by a separate, more recent UK trial (Nature Medicine, 2025), which compared minimally processed and ultra-processed diets that both followed the same official healthy-eating guidance. That trial also found more weight loss on the minimally processed diet, but it drew a direct scientific critique: a group of Harvard and University of Pennsylvania researchers argued the two diets weren’t as nutritionally matched as claimed, and that projecting the 8-week weight difference out to a full year overstated the likely real-world impact. That critique was about this separate UK trial, not Hall’s original 2019 study, but it’s a useful reminder that even well-designed controlled trials in this field can be read differently by serious scientists, and that headline weight differences don’t always translate into large long-term effects.

The wider policy picture

This research hasn’t stayed confined to journals. The finalised 2025–2030 Dietary Guidelines for Americans, published in January 2026, went further than any previous edition in advising people to limit “highly processed foods” – notable, given the guidelines’ own advisory committee had concluded a year earlier that evidence and definitions around “ultra-processed foods” were too inconsistent for firm advice. Tellingly, the final guidelines used “highly-processed” rather than adopting “ultra-processed” as an official category – a sign the underlying science, including how to even define these foods, is still unsettled.

The politics around this research became personal in April 2025, when Kevin Hall announced he was leaving the NIH after 21 years. He said he had experienced censorship in how his ultra-processed food research was reported, alleging that officials linked to the Department of Health and Human Services blocked him from speaking to journalists about newer findings on food and addiction, and that hiring freezes hampered his work. HHS disputed his account. Whatever the rights and wrongs of that dispute, it’s a reminder that nutrition research on a topic this commercially and politically sensitive doesn’t happen in a vacuum – which is exactly why independent replication and transparent methods matter so much here.

The takeaway

The 2019 NIH trial remains the strongest piece of direct, causal evidence that ultra-processed food can drive people to eat more and gain weight, even when a diet looks matched on paper. That’s genuinely rare in nutrition science, where most claims rest on observational data that can’t rule out reverse causation or confounding. But one trial of 20 people, however well designed, is a starting point rather than a final word – the exact mechanism is still being worked out, a separate 2025 trial has faced its own methodological pushback, and official guidance is still feeling its way toward a settled definition of “ultra-processed” at all.

The practical, low-risk takeaway holds regardless: build meals around minimally processed foods where you reasonably can, and treat convenience foods as a matter of degree rather than an all-or-nothing rule. If you want to go deeper into this research, Chris van Tulleken’s Ultra-Processed People covers the science, including Hall’s own work, in far more depth than we can here. This is general information, not medical advice – if you have concerns about your weight or eating patterns, it’s worth discussing them with your GP or a registered dietitian.

FAQs

What is the ultra-processed food weight gain study?

It’s a 2019 randomised controlled trial run by NIH researcher Kevin Hall, published in Cell Metabolism, which found that adults given free access to an ultra-processed diet ate around 508 more calories a day and gained weight, compared with a calorie- and nutrient-matched unprocessed diet.

How many people were in the study?

Twenty adults took part, each spending two weeks on the ultra-processed diet and two weeks on the unprocessed diet, in random order, while living at the NIH Clinical Center.

Why is this trial considered stronger evidence than other UPF studies?

Because it was a controlled, randomised, crossover experiment rather than an observational study. The two diets were matched for calories, sugar, sodium, fibre, and macronutrients, so the extra eating on the ultra-processed diet couldn’t easily be explained by differences in nutrient content alone.

Does this prove ultra-processed food causes obesity?

It provides strong evidence that ultra-processed food can cause people to eat more and gain weight over a short period, in a controlled setting. It’s one trial of 20 people over two weeks per diet, so it can’t tell you everything about long-term, real-world obesity risk on its own.

What causes people to overeat on ultra-processed diets?

The exact mechanism is still debated. A 2025 analysis of the trial data pointed to faster eating rate and higher energy density as likely drivers, alongside the food’s palatability, but researchers haven’t fully separated these factors.

Do the 2025 Dietary Guidelines for Americans tell you to avoid ultra-processed food?

The finalised 2025–2030 guidelines advise limiting “highly-processed foods,” which is new territory for US dietary guidance. However, they use that term rather than “ultra-processed foods,” reflecting ongoing disagreement over how to define the category.

Why did Kevin Hall leave the NIH?

Hall announced in April 2025 that he was leaving after 21 years, saying he experienced censorship in how his ultra-processed food research was reported and that agency hiring freezes hampered his work. NIH’s parent department disputed his characterisation of events.

Has anyone criticised this research?

A separate 2025 UK trial comparing minimally processed and ultra-processed diets drew a published critique from Harvard and University of Pennsylvania researchers, who argued its findings were over-extrapolated and that the diets may not have been as nutritionally matched as claimed. That critique concerned the newer UK trial, not Hall’s original 2019 study.

Is all processed food bad for you?

No. “Processed” covers a huge range, from tinned beans and wholemeal bread to fizzy drinks and packaged pastries. The research on harm concentrates on the most heavily formulated, ready-to-eat products, not on processing in general.

What should I actually do with this information?

Building meals around less-processed staples where practical is a reasonable, low-risk approach supported by this evidence. This is general information, not medical advice – speak to your GP or a registered dietitian about your own health and diet.