Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake
Hall et al. housed 20 weight-stable adults (10 men, 10 women) on a metabolic ward for 28 days and gave them, in random order, 2 weeks of an ultra-processed diet and 2 weeks of an unprocessed diet. The meals offered were matched for calories, energy density, macronutrients, sugar, sodium and fibre, and participants ate as much or as little as they wanted.
On the ultra-processed diet they ate 508 ± 106 kcal/d more (p = 0.0001), from carbohydrate and fat but not protein, and gained 0.9 ± 0.3 kg; on the unprocessed diet they lost 0.9 ± 0.3 kg. They did not rate the ultra-processed meals as more pleasant or familiar, but ate them faster. In respiratory chambers, sleeping and sedentary expenditure did not differ between diets.
The authors also note that the calorie figures probably overstated the unprocessed, high-fibre diet's metabolizable energy (their estimate ~330 kcal/d, not measured).
Protocol
n = 20 weight-stable adults (10 male, 10 female), aged 31.2 ± 1.6 y, BMI 27 ± 1.5 (mean ± SE); inpatients at the NIH Clinical Center for a continuous 28 days.
Randomised to the ultra-processed or unprocessed diet (NOVA classification) for 2 weeks, then immediately the other for 2 weeks; no run-in, no washout.
Three daily meals on a 7-day menu, matched for presented calories, energy density, macronutrients, fibre, sugars and sodium (fibre in the ultra-processed diet partly added via beverages); snacks and water available all day; food offered at twice estimated needs; eat as much or as little as desired; leftovers weighed.
Daily cycle ergometry, three 20-min bouts at 30–40% of heart-rate reserve. One day a week in a respiratory chamber; doubly labelled water over each diet period; body composition by DXA.
Key Results
| Outcome | Ultra-processed vs unprocessed (full text) |
|---|---|
| Ad libitum energy intake | +508 ± 106 kcal/d (p = 0.0001); final week +459 ± 105 kcal/d (p = 0.0003) |
| By macronutrient | carbohydrate +280 ± 54 kcal/d; fat +230 ± 53 kcal/d; protein −2 ± 12 kcal/d (p = 0.85) |
| Body weight | +0.9 ± 0.3 kg (p = 0.009) vs −0.9 ± 0.3 kg (p = 0.007); weight change vs intake difference r = 0.8 |
| Body fat | +0.4 ± 0.1 kg (p = 0.0015) vs −0.3 ± 0.1 kg (p = 0.05) |
| Eating rate | +17 ± 1 kcal/min (p < 0.0001) |
| Pleasantness, familiarity, hunger, fullness | no significant differences |
| 24-h expenditure in chamber | tended higher on ultra-processed, 51 ± 27 kcal/d (p = 0.06); no differences in sleeping or sedentary expenditure, or physical activity |
| Expenditure by doubly labelled water | 171 ± 56 kcal/d higher on ultra-processed (p = 0.006), attributed to the different states of energy balance |
| Fibre eaten | 48.5 ± 2.3 vs 45.8 ± 2.3 g/d (p = 0.41) |
| Authors' estimate of overstated energy in the unprocessed diet | ~330 kcal/d (at ~46 g/d fibre, ~77% insoluble, ~7.2 kcal less per gram of fibre; a calculation, not measured) |
Mechanisms
Measured: faster eating on the ultra-processed diet, moderately correlated with the excess intake (r = 0.45); non-beverage foods ~85% more energy-dense; protein intake held constant while carbohydrate and fat rose.
Authors' interpretation: protein leverage could explain at most ~50% of the excess; softer, faster-eaten food and higher energy density may delay satiety signals. The trial was not designed to identify the cause.
Measured: 24-h chamber expenditure tended higher, not lower, on the ultra-processed diet (51 ± 27 kcal/d, p = 0.06), with no difference in sleeping or sedentary expenditure; the thermic effect of meals was not isolated. The authors contrast this with the ~60 kcal lower post-meal expenditure after processed sandwiches in REF-1129.
Authors' interpretation: the energy balance on the unprocessed diet did not add up (intake minus expenditure was 382 ± 92 kcal/d above the measured loss of body energy), which they attribute either to chance or to its metabolizable energy being overestimated because of its insoluble fibre; faeces were not collected.
Practical Application
Processing changed how much people ate: with matched menus and free choice of portion, 20 adults ate 508 ± 106 kcal/d more on ultra-processed food and gained 0.9 kg in 2 weeks, while losing 0.9 kg on unprocessed food (B+), without finding the ultra-processed meals more pleasant. Absorption was not measured.On the ultra-processed diet, intake minus expenditure matched the measured gain in body energy. Transfer, not tested in athletes: a mostly unprocessed diet may make intake easier to control when holding or losing weight.
Calorie counts may overstate a very high-fibre unprocessed diet: the authors estimate ~330 kcal/d at ~46 g/d of fibre (calculation, not measured); REF-1131 measured 116 kcal/day of extra faecal loss on a designed high-fibre diet. Either way the body-mass trend, not the count, is the arbiter (SYNTH-87).
The flip side for heavy training days: the same satiety that helps a cut makes it hard to eat enough when the day's load is high; meeting carbohydrate targets (SYNTH-69) can justify energy-dense, more processed carbohydrate sources (inference, not tested here).
No thermic-effect argument for whole food: chamber expenditure was not lower on the ultra-processed diet (24-h tended higher, p = 0.06; sleeping and sedentary expenditure did not differ).
Limitations
n = 20, 2 weeks per diet, no run-in or washout; ward setting, food free and chosen for participants, so cost and convenience did not operate.
The diets were matched on presented nutrients; real-world ultra-processed diets differ more in sugar, fat and fibre, so the effect may be larger or different outside the ward.
DXA in short trials is affected by fluid shifts (sodium intake differed: 5.8 vs 4.6 g/d).
Faeces were not collected, so metabolizable energy was calculated from databases, not measured; the ~330 kcal/d overestimate is the authors' calculation.
General population with light mandated cycling; athletes' appetite under heavy training was not studied.
How this entry was checked and changed
Why this level
randomised crossover, n = 20 (rubric: crossover ≥15 = B+), metabolic ward with weighed intake, respiratory chambers and doubly labelled water; read from the full text (author manuscript)
Full text (checked 2026-10-13)
10 male and 10 female weight-stable adults aged 31.2±1.6 y, BMI=27±1.5 kg/m2; continuous 28-day inpatient stay (full text).
Energy intake 508±106 kcal/d greater during the ultra-processed diet (p=0.0001); 459±105 kcal/d greater in the final week (p=0.0003) (full text).
Weight +0.9±0.3 kg (p=0.009) vs −0.9±0.3 kg (p=0.007); body fat +0.4±0.1 kg vs −0.3±0.1 kg (full text).
Eating rate 17±1 kcal/min greater on the ultra-processed diet (p<0.0001) (full text).
Chamber 24-h expenditure 51±27 kcal/d higher on the ultra-processed diet (p=0.06); no differences in sleeping energy expenditure, sedentary energy expenditure or physical activity (full text).
With ~46 g/d of fibre, ~77% insoluble, the expected decrease in metabolizable energy "amounts to ~330 kcal/d" (authors' calculation) (full text).