This is not medical advice. General information about food tracking, not a treatment plan. If you are under 18, pregnant or breastfeeding, have ever had an eating disorder or a difficult relationship with food, or are managing a medical condition, talk to a doctor or a registered dietitian before you start counting calories or change your intake. The full list is here.

What ultra-processed means, and who decides

About 57% of the calories US adults eat come from ultra-processed food. The figure is from 24-hour dietary recalls of 40,937 adults in the NHANES surveys, and it rose from 53.5% in 2001 to 57.0% in 2018.1 So this is most of what most people log, me included.

The term comes from Nova, a classification built at the University of São Paulo that sorts food into four groups by the extent and purpose of industrial processing. Group four, ultra-processed, covers products assembled from substances split out of whole foods and then modified, usually with additives that change flavour, colour or texture. The practical test its authors give is the ingredient list. A product counts if it contains at least one substance "never or rarely used in kitchens", such as high-fructose corn syrup, hydrogenated oils or hydrolysed proteins, or an additive whose job is to make the product more appealing, such as flavours, colours, emulsifiers or sweeteners.2

Calories, fat, sugar and protein do not appear anywhere in that definition. A calorie and nutrient tracker records exactly those four things, so two foods it scores identically can sit in different Nova groups.

508 calories a day, from meals with the same numbers

In 2019 Kevin Hall's group at the US National Institutes of Health admitted 20 weight-stable adults, average age 31 and average BMI 27, to a clinical ward for four weeks. Each person ate an ultra-processed diet for two weeks and an unprocessed one for two weeks, in random order. The meals on the two diets were designed to be matched for the calories presented, energy density, macronutrients, sugar, sodium and fibre, and people were told to eat as much or as little as they wanted.3

On the ultra-processed diet they ate 508 calories a day more, give or take 106. The extra came as carbohydrate (280 calories) and fat (230 calories), and protein intake did not change. They gained 0.9 kg in the ultra-processed fortnight and lost 0.9 kg in the unprocessed one.3

Two details matter if you count calories in food. The first is that weight followed intake: the correlation between energy eaten and weight change was 0.8.3 The trial gives no sign that an ultra-processed calorie weighs more than any other. People just ate more of them. The second is that a nutrition panel would not have warned anyone. The trays were built to carry matching numbers, and the gap opened in how much of each tray got eaten.

It was also twenty people, two weeks per diet, living on a hospital ward and eating food that somebody else chose and cooked. The authors hedge their conclusion to match: limiting ultra-processed food "may be" an effective strategy.3

Why people ate more: speed and density

Hall's design matched the nutrients on offer, so the cause has to be something else. The follow-up work points at two properties of the food.

Fazzino and colleagues, with Hall, pooled 2,733 meals from feeding studies covering four dietary patterns and asked which features of a meal went with eating more of it. Energy density (calories per gram), eating rate and hyper-palatable foods were each positively related to the energy eaten at a meal, on all four diets.4

A group in Singapore tested texture directly. Fifty healthy-weight adults with an average age of 24 ate four lunches matched for the energy served: soft or hard in texture, minimally processed or ultra-processed. The hard-textured lunches were eaten more slowly, and people took in 26% less energy from them at either processing level. The smallest lunch was the hard minimally processed one at 483 calories and the largest was the soft ultra-processed one at 789.5

A Dutch trial then kept the processing and changed the rest. Sixty-nine adults ate sandwiches made with 67 to 98% ultra-processed ingredients by energy, built to differ in energy density and in how fast they could be eaten. People ate 570 calories of the slow, low-density sandwich and 1,143 of the fast, dense one.6 Both sandwiches were ultra-processed, and one meal came out half the size of the other. Two of that paper's authors declare speaking fees or travel money from food companies, including Nestlé, PepsiCo and Mondelez, and the paper says so.6

These are single lunches eaten in a lab by young, mostly lean adults. They tell you about one meal and nothing about a year.

When both diets followed the guidelines

Hall's ultra-processed menu matched the other one on nutrients, but nobody would call it a recommended diet. In 2025 a trial in England asked what happens when both diets follow national healthy-eating guidance.

Fifty-five adults with a BMI between 25 and 40, who normally got at least half their calories from ultra-processed food, were given two eight-week diets in random order. Both followed the UK's Eatwell Guide and both were eat-as-much-as-you-like. One was made of minimally processed food and the other of ultra-processed food. Fifty people made up the main analysis.7

People lost weight on both. The minimally processed diet produced a 2.06% loss over eight weeks and the ultra-processed diet 1.05%, a difference of about one percentage point (95% CI 0.14 to 1.87).7 By my arithmetic, for someone who starts at 90 kg that is roughly 1.9 kg against 0.9 kg.

I read that two ways. An ultra-processed diet built to healthy guidelines still led to weight loss, without anyone being told to eat less, in people whose usual diet was mostly ultra-processed. And processing still cost about half the loss. The confidence interval runs from almost nothing to nearly two points, so the size of that cost is loosely pinned down.

The cohort studies, and how much weight they can carry

Most of what you read about ultra-processed food and health comes from cohort studies, which follow large groups for years and compare those who eat more of it with those who eat less. A 2024 umbrella review in the BMJ collected 45 pooled analyses covering close to 9.9 million participants. Higher exposure was associated with a worse outcome in 32 of the 45, obesity among them (odds ratio 1.55, 95% CI 1.36 to 1.77).8

The same review graded its own evidence, and the grades rarely get quoted. Of the 45 analyses, 22 were rated low quality, 19 very low and four moderate, which leaves none rated high. The obesity finding was graded low.8

Robinson and Jones looked at how fragile the weight-gain association is. Pooling the prospective studies gave a relative risk of 1.14 for weight gain with higher intake. They then worked out how strong an unmeasured confounder would need to be to produce that figure on its own, and the answer was small to moderate. Their examples were depressive symptoms, a tendency to overeat and access to healthy food. Unaccounted-for confounding, they concluded, "could plausibly explain" the association.9

The association may still be real. The cohort studies cannot show it either way, which leaves a handful of small feeding trials carrying most of the argument.

Sweetened drinks and wholegrain bread share a category

Nova's fourth group is wide. An analysis of three large US cohorts, about 207,000 health professionals in all, split ultra-processed food into its groups and followed cardiovascular disease. The fifth of people eating the most had an 11% higher risk than the fifth eating the least (hazard ratio 1.11, 95% CI 1.06 to 1.16). Inside that total the groups pointed in different directions. Sweetened drinks, with sugar or without, and processed meats were associated with higher risk. Bread and cold cereals, yoghurt and dairy desserts, and savoury snacks were associated with lower risk.10 The authors note that 95 to 98% of participants were white and ask for replication in more diverse populations.10

I would not eat crisps on the strength of that. Associations among health professionals filling in food questionnaires are the kind of evidence the last section marked down. It does show that "ultra-processed" is a poor guide to any single item on a plate.

What the evidence does not settle

Start with the category itself. A French team asked food and nutrition specialists to sort foods into the four Nova groups: 120 marketed products with their ingredient lists, and 111 generic foods without. Agreement between evaluators, measured by Fleiss' kappa where 1 is perfect, was 0.32 and 0.34. The authors concluded that the current criteria "do not allow for robust and functional food assignments".11 If specialists with the ingredients in front of them disagree, shoppers and apps will too.

The cause is open as well. Energy density and eating rate explain a good deal in single meals,456 but none of the trials I have cited isolated the additives the definition is built around, and none showed that speed and density account for the whole effect over months.

Then there is duration. The longest feeding trial here ran eight weeks per diet.7 Whether a 500-calorie gap holds or fades over a year has not been tested in a trial I could find. Hall's participants also did not shop for or cook any of their food, and convenience is a large part of why people eat this way outside a ward.

What a calorie and nutrient tracker can see, and what it cannot

MyPlate returns calories, protein, carbohydrate and fat for each dish you log from a photo or a sentence, along with the ingredients it estimated. It has no field for processing and I am not planning one. It has no barcode scanner and no food database, so it never reads an ingredient list, and the ingredient list is where Nova says the call has to be made.2 A photo of a chicken sandwich cannot show whether the bread contains an emulsifier. Given how the specialists scored, I would not trust a label generated that way.11

What a nutrition food tracker does see is the result. In Hall's trial the processing showed up as 508 more calories eaten, and calories are what a log records.3 If ultra-processed meals lead you to eat more, a complete log shows it as days over target. The weak point is "complete". The extra tends to arrive as a second helping or a bag finished at the desk, and those are easy entries to skip. The post on how accurate calorie tracking apps are covers how much people leave out.

A log also shows a pattern. Scroll back over a week and you can see which meals ran past what you planned. That is closer to your own diet than a category that files cola next to wholegrain bread. The numbers themselves carry a smaller bias in the same direction, because whole nuts and whole grains deliver somewhat less energy than their labels say. That one is in the post on the calories you actually absorb.

How to use this if you count calories

  1. Keep counting. In the inpatient trial weight followed the calories eaten,3 so the total in your log is still the number that moves the scale.
  2. When a day runs over, look at which meals did it. If they were the soft, dense, quick ones, that fits what the lab lunches found,456 and it tells you where to make a change.
  3. Change the food before you cut the target. Something you have to chew, or something with more bulk for the same calories, is what those studies tested. Whatever you swap, keep the day above 1,200 calories for women and 1,500 for men, the floor in the NIH clinical guidelines. MyPlate will not set a target below it.12
  4. Leave the bread alone if it is working for you. The cohort signal sat with sweetened drinks and processed meat, and bread, cereal and yoghurt pointed the other way.10
  5. If you want to check a product, read the ingredient list for things you would not find in a kitchen,2 and hold the answer loosely. Specialists given the same list often disagreed.11

I build an app for people who do not have a minute to log a meal. Telling them to cook everything from scratch would be an odd thing for me to do, and the English trial suggests it is not required, since the ultra-processed diet that followed the guidelines still produced weight loss.7

And a limit. If sorting food into good and bad has started to make eating feel like a test, or a number on a screen is deciding how you feel about yourself, stop and talk to a doctor or a registered dietitian. The warning signs worth knowing are in the post on calorie tracking and eating disorders.

Common questions

Do calories from ultra-processed food count the same?

For body weight, the best trial so far says yes. In a two-week inpatient study of 20 adults, weight change tracked energy intake closely (r = 0.8), and people gained on the ultra-processed diet because they ate 508 calories a day more of it, from meals matched for the nutrients on offer.3 That is one small, short trial, so it does not rule out a smaller effect of processing that it could not detect.

Can you lose weight eating ultra-processed food?

Yes, in the one trial that tested it under healthy-eating guidelines. Fifty adults in England lost 1.05% of their weight over eight weeks on an ultra-processed diet that followed the UK Eatwell Guide, eating as much as they liked. They lost 2.06% on a minimally processed diet built to the same guide, so processing roughly halved the loss.7 Eight weeks is the longest this has been tested.

Is ultra-processed food bad for you?

Higher intake is associated with worse health in cohort studies, but most of that evidence is weak. A 2024 umbrella review found associations with 32 of 45 outcomes and rated 41 of the 45 analyses low or very low quality.8 A separate analysis concluded that unmeasured confounding could plausibly explain the link with weight gain.9 The associations also differ by food: sweetened drinks and processed meats carried the higher cardiovascular risk in three US cohorts, and bread, cereals and yoghurt did not.10

How do I know if a food is ultra-processed?

Read the ingredient list. Under the Nova system a product is ultra-processed if it contains at least one substance rarely used in home kitchens, such as high-fructose corn syrup, hydrogenated oils or hydrolysed proteins, or additives such as flavours, colours, emulsifiers and sweeteners.2 The rule is less clear-cut than it sounds. When French food specialists classified 120 products with their ingredients listed, agreement between them was low, with a Fleiss' kappa of 0.32.11

Can a calorie counter app tell you if a food is ultra-processed?

Not reliably. Processing is defined by the ingredient list, and calories and macros say nothing about it.2 An app that scans barcodes could read the ingredients and apply the Nova rule, but specialists applying that rule by hand often disagreed with each other.11 MyPlate does not try. It has no barcode scanner and no food database, and it estimates calories and macros from a photo or a description. What a food tracker for weight loss can show is whether you ate more, which is the effect the trials measured.3

Count the calories, then look at the food

Ultra-processed food is defined by ingredients, and a calorie and nutrient tracker never records those. In the two feeding trials that tested it, people ate more or lost less on the ultra-processed diet, and in the one that reported it their weight followed the calories. So the log still does its job of telling you how much. It cannot tell you why a meal was easy to overeat. For that, look at which entries keep running over, and treat "ultra-processed" as a rough hint. The people who pooled the cohort evidence rated 41 of their 45 analyses low or very low quality.

Sources

Every figure above traces to one of these. If you find a number that doesn't match the source it claims, tell me and I'll correct it.

  1. Juul F, Parekh N, Martinez-Steele E, Monteiro CA, Chang VW (2022). Ultra-processed food consumption among US adults from 2001 to 2018. American Journal of Clinical Nutrition, 115(1):211-221. Read on PubMed →
  2. Monteiro CA, Cannon G, Levy RB, et al. (2019). Ultra-processed foods: what they are and how to identify them. Public Health Nutrition, 22(5):936-941. Read on PubMed →
  3. Hall KD, Ayuketah A, Brychta R, et al. (2019). Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metabolism, 30(1):67-77.e3. Read on PubMed →
  4. Fazzino TL, Courville AB, Guo J, Hall KD (2023). Ad libitum meal energy intake is positively influenced by energy density, eating rate and hyper-palatable food across four dietary patterns. Nature Food, 4(2):144-147. Read on PubMed →
  5. Teo PS, Lim AJ, Goh AT, et al. (2022). Texture-based differences in eating rate influence energy intake for minimally processed and ultra-processed meals. American Journal of Clinical Nutrition, 116(1):244-254. Read on PubMed →
  6. Lasschuijt MP, Heuven LA, Gonzalez-Estanol K, Siebelink E, Chen Y, Forde CG (2025). The effect of energy density and eating rate on ad libitum energy intake in healthy adults: a randomized controlled study. Journal of Nutrition, 155(8):2602-2610. Read on PubMed →
  7. Dicken SJ, Jassil FC, Brown A, et al. (2025). Ultraprocessed or minimally processed diets following healthy dietary guidelines on weight and cardiometabolic health: a randomized, crossover trial. Nature Medicine, 31(10):3297-3308. Read on PubMed →
  8. Lane MM, Gamage E, Du S, et al. (2024). Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses. BMJ, 384:e077310. Read on PubMed →
  9. Robinson E, Jones A (2024). Causality or confounding? Applying E values to examine associations between ultra-processed food consumption and risk of weight gain. International Journal of Obesity, 48(9):1342-1346. Read on PubMed →
  10. Mendoza K, Smith-Warner SA, Rossato SL, et al. (2024). Ultra-processed foods and cardiovascular disease: analysis of three large US prospective cohorts and a systematic review and meta-analysis of prospective cohort studies. Lancet Regional Health Americas, 37:100859. Read on PubMed →
  11. Braesco V, Souchon I, Sauvant P, et al. (2022). Ultra-processed foods: how functional is the NOVA system? European Journal of Clinical Nutrition, 76(9):1245-1253. Read on PubMed →
  12. National Heart, Lung, and Blood Institute, NIH (1998). Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults. NIH Publication 98-4083. Read the guidelines →

The formulas and limits behind MyPlate's own numbers, with their citations, are on the health sources page.