Ultra-processed food: does bread, yogurt or cereal count?
Start with the four processing groups, then look at familiar foods.
On this page
Ultra-processed foods (UPFs) are industrial formulations classified as group 4 in NOVA, a system based on the extent and purpose of food processing. They commonly combine substances extracted from foods with ingredients used to alter flavor, texture, or appearance.[1]
What does this actually mean?
The meaning
Ultra-processed food describes a way of making and formulating food, rather than simply putting it in a package. In NOVA, group 1 covers foods such as oats, fruit and plain milk, including minimal preservation; group 2 covers cooking ingredients such as oil, salt and sugar; group 3 combines foods with those ingredients, as in simple bread; and group 4 covers industrial formulations, often using extracted food substances and ingredients that shape flavor or texture. The group numbers classify processing. They are not four levels of nutritional quality.[22]
Familiar examples
Bread, yogurt and cereal are families of products, not fixed processing categories. A simple bread recipe and an industrial formulation can differ. Plain yogurt made from milk and cultures is minimally processed; a sweetened, colored and flavored formulation can be ultra-processed. Plain oats also differ from a ready-to-eat cereal with flavorings and other industrial ingredients. Look at the actual recipe rather than treating a loaf, a plastic tub or a cardboard box as evidence by itself.[23]
What it does not tell you
A processing category is not the same as a nutrient score, and NOVA is not the only classification system. Other systems draw boundaries differently or incorporate nutrient criteria. That disagreement matters when comparing studies or food-rating apps: first check whether they are classifying the same thing. It does not mean that nutrition information disappears when a processing assignment is uncertain.[16]
What the evidence supports
In a 2025 trial of adults with overweight or obesity, both eight-week menus followed dietary guidance. Weight fell on both, more on the minimally processed menu. This supports a difference between those menus; it does not measure the lifetime cancer risk of your breakfast cereal.[3]
Why the whole food still matters
A separate meal experiment changed eating rate and calorie density. Participants ate less energy from the slower, lower-density meal. That helps explain why texture and the calories packed into each gram can matter alongside processing. A lunch experiment cannot establish that every product carrying the same processing label has identical long-term health effects.[13]
Where uncertainty remains
The FDA describes unanswered questions about how and why these foods may affect health and an effort to develop a uniform US definition. Its public page documents that work; it does not turn every NOVA assignment into a legal US labeling category. In practical terms, a research classification, a government policy definition and an individual product assessment can answer different questions. Check which definition a strong headline is using.[24]
What to check when shopping
Read the ingredients and compare foods you would actually substitute for one another. Alongside processing, consider how much wholegrain or other fiber-rich food a choice contributes, and compare its nutrient panel rather than the front slogan alone. Convenient packaged foods can still help assemble meals. The useful shopping question is what this choice adds to, or replaces in, your usual pattern of vegetables, fruit, beans and wholegrains.[15][25]
What does “processed” actually mean?
Freezing and pasteurization can remain minimal processing in NOVA when other food substances are not added. The formulation and purpose matter.[1]
Processing can also serve food safety. For example, EFSA describes high-pressure processing as a preservation technique that uses pressure to reduce microorganisms in foods such as juices or ready-to-eat products. A processing method can have a useful purpose without telling us everything about a food’s nutritional value.[10]
Ingredients have different jobs, too. FDA distinguishes preservatives that slow spoilage, added nutrients that restore or improve nutritional value, and ingredients that affect flavor or texture. An additive’s presence is not, by itself, evidence that the finished food causes disease; ingredient safety and whole-diet health effects are different questions.[9]
The four NOVA groups
NOVA separates foods, cooking ingredients, and industrial formulations. The examples below illustrate the categories; the exact recipe and manufacturing information can change a product’s assignment.[4][7]
| Group | What it means | Typical examples |
|---|---|---|
| 1 · Unprocessed or minimally processed | Foods from plants or animals, including simple preservation or preparation. | Fruit, vegetables, rice, plain milk, eggs, unsalted nuts.[4] |
| 2 · Processed culinary ingredients | Substances used to season, cook, or prepare other foods. | Vegetable oils, butter, sugar, salt.[4] |
| 3 · Processed foods | Group 1 foods combined with cooking ingredients, often for preservation. | Vegetables canned with salt, traditional cheese, simple bread.[4] |
| 4 · Ultra-processed foods | Industrial formulations using multiple ingredients and manufacturing techniques. | Soft drinks, packaged snack products, instant noodles, many confectionery products.[4] |
These numbers describe processing categories, not a four-point nutrition score. Sugar and salt belong to group 2; their classification does not imply unlimited consumption is healthy.[5]
Everyday examples: similar foods, different groups
| Food | How the formulation changes its category |
|---|---|
| Bread | Traditional artisanal or home-baked recipes: group 3. Some industrial formulations containing emulsifiers: group 4.[11] |
| Breakfast cereal | Plain oats: group 1. Oats with sugar: group 3. Formulations with flavorings or colors: group 4.[1] |
Food questionnaires may record only “bread” without these recipe details. A 2025 critical review highlights how that can blur processing categories in observational studies. These examples explain classification; they do not establish the health effect of each product.[11]
NOVA is one classification system, not a universal laboratory measurement
A 2025 review identified six systems: NOVA, EPIC, IFPRI, UNC, UP3 and Siga. Their definitions of highly or ultra-processed foods differ. Some emphasize manufacturing purpose and additives; Siga also incorporates quantitative nutrient criteria. Thus the same processing label from two systems may not describe exactly the same exposure. This atlas uses NOVA when it discusses its four groups and specifies the study’s system when interpreting evidence. The review compares definitions; it does not establish that one scoring system predicts every clinical outcome better.[16]
How to recognize a product—and where labels fall short
NOVA’s authors suggest checking for ingredients uncommon in home cooking, such as hydrolysed proteins or modified oils, and additives shaping sensory qualities, such as flavorings, colors, or emulsifiers. This is a classification clue, not a toxicology test; preservatives alone do not establish ultra-processing.[1]
Some cases are ambiguous. In a 2022 study, French food and nutrition specialists assigned products to NOVA groups with low overall agreement, even when ingredient lists were supplied. That does not make every assignment impossible; it shows why an article should explain its coding rules and uncertainty rather than present every product label as definitive.[7]
When formulation details are missing, an uncertain classification is more informative than guessing a group. Classification studies help identify where the available information cannot support a confident assignment.[7]
Processing and nutritional quality are related, but different
Processing categories describe manufacture; nutritional assessment examines energy, fiber, salt, sugars, or saturated fat. SACN’s 2025 review found partial overlap between ultra-processing and unfavorable nutrient profiles, with subgroup differences. Both need interpretation alongside the overall dietary pattern.[5]
Fortification is another example of the distinction. Adding vitamins or minerals can improve nutritional value, while a food may also contain other ingredients that influence its processing classification. “Fortified,” “processed,” and “nutritious” answer different questions; one label does not substitute for the others.[9]
Evidence & uncertainty
The science is incomplete. Feeding trials support effects of particular ultra-processed menus on intake or weight, while long-term disease findings largely come from observational research. It remains uncertain how much risk comes from processing itself, nutritional composition, or other differences between diets and people.[5]
Selected-source overview · Independent review pendingPublished questions
What is an ultra-processed food (UPF)?
Selected-source overview · Independent review pending. Read the evidence context.
In NOVA, it is a group 4 industrial formulation, often combining extracted food substances with ingredients that shape flavor, texture, or appearance. Ultra-processing is a processing-and-formulation category, not simply a synonym for packaged food.[1]
Evidence behind this answer
- Ultra-processed foods: what they are and how to identify them
classification commentary · 2019
Selected publisher full-text sections · Defining ultra-processed foods; Non-ultra-processed food groups; Identifying ultra-processed foods
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Are all processed foods ultra-processed?
Selected-source overview · Independent review pending. Read the evidence context.
No. NOVA has four groups. It distinguishes foods and minimally processed foods, cooking ingredients, processed foods, and ultra-processed formulations. Cooking, preservation, or packaging alone does not make every food group 4.[4]
Evidence behind this answer
- 2017/2018 POF: fresh food and culinary preparations prevail in the national diet
official classification explanation · 2020
Indexed official-page text; direct retrieval was blocked · Paragraph describing the four NOVA groups and examples
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
What are the four NOVA food groups?
Selected-source overview · Independent review pending. Read the evidence context.
They are: 1, unprocessed or minimally processed foods; 2, processed culinary ingredients; 3, processed foods; and 4, ultra-processed foods. The comparison table gives their meanings and examples.[4]
Evidence behind this answer
- 2017/2018 POF: fresh food and culinary preparations prevail in the national diet
official classification explanation · 2020
Indexed official-page text; direct retrieval was blocked · Paragraph describing the four NOVA groups and examples
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Can ultra-processed diets make people eat more calories?
Selected-source overview · Independent review pending. Read the evidence context.
Yes, particular tested menus can. In the small 2019 inpatient trial, adults ate about 508 calories a day more on the ultra-processed menu. The result cannot be assigned to every individual product.[2]
Evidence behind this answer
- Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake
randomized crossover trial · 2019
Indexed abstract and figure captions · Abstract; Figure 2: food intake and eating rate
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does an additive mean an ultra-processed food is harmful?
Selected-source overview · Independent review pending. Read the evidence context.
An additive alone does not establish harm. Ingredients serve different functions, and their safety under particular uses is a different question from the long-term effects of eating a whole dietary pattern.[9]
Evidence behind this answer
- Food Additives and GRAS Ingredients — Information for Consumers
agency explanation · Date not stated
Official page text · Functions of ingredients; Safety standard and oversight
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Do ultra-processed food studies prove that processing causes disease?
Selected-source overview · Independent review pending. Read the evidence context.
Not across all diseases or foods. Feeding trials support effects of specific menus on intake or weight; long-term disease associations are mainly observational and have important certainty limits.[6]
Evidence behind this answer
- Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses
umbrella review of observational studies · 2024
Indexed primary-article abstract and text; direct retrieval unavailable · Methods; Results: GRADE certainty; Discussion: bias and confounding
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Can ordinary packaged foods belong to different processing groups?
Selected-source overview · Independent review pending. Read the evidence context.
NOVA groups differ. Frozen peas and plain yoghurt can be minimally processed; simple bread can be processed, while industrial breads and soft drinks can be ultra-processed. Formulation matters, and classification can be uncertain across specific recipes.[1][7]
Evidence behind this answer
- Ultra-processed foods: what they are and how to identify them
classification commentary · 2019
Selected publisher full-text sections · Defining ultra-processed foods; Non-ultra-processed food groups; Identifying ultra-processed foods - Ultra-processed foods: how functional is the NOVA system?
classification reliability study · 2022
Abstract and disclosure statement · Abstract: methods, agreement, conclusions
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does a NOVA label tell the whole story about whether a food is nutritious?
Selected-source overview · Independent review pending. Read the evidence context.
No. NOVA classifies processing rather than giving a complete nutrient profile. Foods in one category can differ in fibre, salt, sugar, energy and other characteristics. Category-level associations therefore do not establish that every product has the same effect. Processing evidence and nutritional composition both matter, alongside remaining uncertainty about causation.[5]
Evidence behind this answer
- Processed foods and health: SACN’s rapid evidence update summary
official evidence assessment · 2025
Official summary text · Results; Limitations; Discussion; Conclusions
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Can two ultra-processed menus produce different calorie intake?
Selected-source overview · Independent review pending. Read the evidence context.
Yes. The texture trial compared two UPF menus and found different intake despite their shared category.[12]
Evidence behind this answer
- Eating rate has sustained effects on energy intake from ultraprocessed diets: a 2-week ad libitum dietary randomized controlled crossover trial
randomized crossover trial · 2026
Original indexed abstract and selected original journal result passages inspected; direct PMC access showed a verification wall · Abstract: two 14-day UPF menus, eating rate, intake and body weight
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Did lower intake in the UPF texture trial guarantee lower body weight?
Selected-source overview · Independent review pending. Read the evidence context.
The trial found lower intake but no clear body-weight difference between its short menu periods.[12]
Evidence behind this answer
- Eating rate has sustained effects on energy intake from ultraprocessed diets: a 2-week ad libitum dietary randomized controlled crossover trial
randomized crossover trial · 2026
Original indexed abstract and selected original journal result passages inspected; direct PMC access showed a verification wall · Abstract: two 14-day UPF menus, eating rate, intake and body weight
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Are eating rate and energy density the same thing?
Selected-source overview · Independent review pending. Read the evidence context.
One describes speed of eating; the other describes energy per weight of food. Both can affect intake.[13]
Evidence behind this answer
- The Effect of Energy Density and Eating Rate on Ad Libitum Energy Intake in Healthy Adults—A Randomized Controlled Study
randomized crossover meal experiment · 2025
Original indexed abstract and relevant passages inspected; direct retrieval unavailable · Abstract, five lunch conditions and conflict-of-interest disclosure
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does a lunchtime UPF experiment establish lifelong disease risk?
Selected-source overview · Independent review pending. Read the evidence context.
A short meal comparison measures intake, not decades of disease outcomes.[13]
Evidence behind this answer
- The Effect of Energy Density and Eating Rate on Ad Libitum Energy Intake in Healthy Adults—A Randomized Controlled Study
randomized crossover meal experiment · 2025
Original indexed abstract and relevant passages inspected; direct retrieval unavailable · Abstract, five lunch conditions and conflict-of-interest disclosure
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Why might eating lots of ultra-processed foods be bad for health?
Selected-source overview · Independent review pending. Read the evidence context.
Energy-dense, convenient food patterns can make excess calorie intake easier and promote weight gain. This mechanism does not establish identical harm from every product in the category.[14]
Evidence behind this answer
- Limit fast foods: recommendation evidence
professional evidence assessment · Date not stated
Relevant original organization assessment passages inspected; no independent synthesis of every underlying study · Original energy-intake, food-pattern, portion and weight-gain explanation; definition differs from NOVA
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does the health question also depend on what these foods replace?
Selected-source overview · Independent review pending. Read the evidence context.
Yes. Replacing fibre- and nutrient-rich foods can change the overall dietary pattern; processing labels alone do not describe that comparison.[15]
Evidence behind this answer
- Wholegrains, vegetables, fruit and beans: recommendation evidence
professional evidence assessment · Date not stated
Relevant original assessment passages inspected; selected dietary-pattern explanation, not individual product appraisal · Original dietary fibre/nutrient and food-replacement interpretation
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Is NOVA the only food-processing classification?
Selected-source overview · Independent review pending. Read the evidence context.
No. Other systems include EPIC, IFPRI, UNC, UP 3 and Siga. Definitions and nutrient criteria differ, so categories cannot always be substituted directly across studies.[16]
Evidence behind this answer
- Definitions of ultra-processed foods beyond NOVA: a systematic review and evaluation
systematic classification review · 2025
Selected original source passages inspected · Abstract: objectives, methods, identified classification systems and comparison limitations
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does disagreement about NOVA mean the feeding trials tell us nothing?
Selected-source overview · Independent review pending. Read the evidence context.
No. A randomized comparison can support effects of its tested menus even while researchers dispute whether a broad processing category captures an independent causal factor or the best policy target.[18]
Evidence behind this answer
- The industrial formulation fallacy: a critical appraisal of the Nova classification system
critical narrative analysis · 2026
Original abstract inspected; author interpretation, not a definitive causal assessment · Abstract: causal critique, structural determinants and policy interpretation
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Is all supermarket bread ultra-processed?
Selected-source overview · Independent review pending. Read the evidence context.
No. Simple breads and industrial formulations differ. Check the ingredients and formulation; packaging or supermarket sale alone does not assign a NOVA group.[22]
Evidence behind this answer
- Ultra-processed foods, diet quality, and health using the NOVA classification system
agency commissioned classification report · 2019
Selected report text, author-uploaded copy inspected; official repository retrieval failed · Section 2, pages 6–12: groups and ingredient examples
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Is plain yogurt ultra-processed because it comes in a plastic tub?
Selected-source overview · Independent review pending. Read the evidence context.
No. The Brazilian guide lists plain yogurt as minimally processed and distinguishes sweetened, colored and flavored formulations. The container alone is not the classification.[23]
Evidence behind this answer
- Dietary Guidelines for the Brazilian Population, second edition
government dietary guidance · 2014
Selected original guideline text inspected through a hosted copy; original ministry URL timed out · Chapter 2; pages 79–80 plain versus formulated yogurts
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does every breakfast cereal belong to the same NOVA group?
Selected-source overview · Independent review pending. Read the evidence context.
No. Plain oats differ from formulations containing flavorings and other industrial ingredients. Classify the particular recipe rather than treating the word cereal as a processing level.[22]
Evidence behind this answer
- Ultra-processed foods, diet quality, and health using the NOVA classification system
agency commissioned classification report · 2019
Selected report text, author-uploaded copy inspected; official repository retrieval failed · Section 2, pages 6–12: groups and ingredient examples
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Does a long ingredient list automatically make bread ultra-processed?
Selected-source overview · Independent review pending. Read the evidence context.
No. NOVA classifies formulation and processing, not an ingredient-count cutoff. Simple bread and industrial formulations can differ; judge the recipe rather than counting ingredients.[22]
Evidence behind this answer
- Ultra-processed foods, diet quality, and health using the NOVA classification system
agency commissioned classification report · 2019
Selected report text, author-uploaded copy inspected; official repository retrieval failed · Section 2, pages 6–12: groups and ingredient examples
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Which breakfast-cereal details help with a shopping comparison?
Selected-source overview · Independent review pending. Read the evidence context.
Check whole-grain ingredients, fibre, sugars and sodium, comparing the same serving amount. Also consider which food this cereal replaces and how it fits your usual dietary pattern.[25][15]
Evidence behind this answer
- Food Labeling Guide, March 2025
government labeling guidance · 2025
Original six-page PDF, selected relevant sections · Pages 2–3: ingredient lists and Nutrition Facts - Wholegrains, vegetables, fruit and beans: recommendation evidence
professional evidence assessment · Date not stated
Relevant original assessment passages inspected; selected dietary-pattern explanation, not individual product appraisal · Original dietary fibre/nutrient and food-replacement interpretation
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Can two studies estimate different UPF intake from the same foods?
Selected-source overview · Independent review pending. Read the evidence context.
Yes. A Chilean study changed its estimated UPF share when researchers used food descriptions, detailed ingredient information or a broader search for additives. A difference in the number can reflect classification choices rather than a change in what someone ate.[26]
Evidence behind this answer
- Impact of the use of food ingredients and additives on the estimation of ultra-processed foods and beverages
original cross sectional methods study · 2023
Original publisher full text: relevant methods, results, limitations, funding and conflict statements read. · Methods2.1–2.4, Results3 and Discussion4: dietary sample, three coding methods, estimates and limitations
This answer summarizes selected sources. An independent appraisal of the complete evidence has not been completed.
Selected-study outcomes
These views show selected findings from abstracts, publication records or selected full-text passages. Access is stated for each record. Complete synthesis and independent appraisal remain pending.
Selected study · review incomplete
UPF in a small feeding trial
Weight and consumed energy increased on the UPF diet.[20]
Nine men, open-label, short duration; fibre and energy density differed.
- Population
- Nine Japanese men with overweight/obesity
- Intervention/exposure
- Ad-libitum UPF diet
- Comparison
- Non-UPF diet; offered energy/macronutrients matched
- Follow-up
- 1 week/diet; 2-week washout
- Setting
- Tokyo inpatient ward
| Measure | Reported result |
|---|---|
| Paired mean difference | 1.1 kg; 95% CI 0.2–2; p 0.021 |
| Sample | 9 participants |
| Secondary energy intake | Paired mean difference: 813.5 kcal/day; 95% CI 342.4–1284.7; p 0.0041 |
| completed follow up | 9 |
Harms: No adverse events observed in this short trial.
Stated population/comparison only.
Extraction details
Access: Indexed primary abstract and publisher methods/results
Funding: Not extracted.
Conflicts: Not extracted.
Query: site:pubmed.ncbi.nlm.nih.gov/39267249 "Methods" "Results"
Certainty ungraded. Appraisal and qualified review pending.
Selected study · review incomplete
UPF and mortality in US cohorts
The adjusted association was small; subgroup foods differed.[21]
Observational confounding, self-reported diet and occupational sampling limit causal inference.
- Population
- 74,563 women and 39,501 men; initially without cancer/CVD/diabetes
- Intervention/exposure
- Highest UPF intake quartile
- Comparison
- Lowest UPF intake quartile
- Follow-up
- Median 34/31 years
- Setting
- US health-professional cohorts
| Measure | Reported result |
|---|---|
| Adjusted hazard ratio | 1.04; 95% CI 1.01–1.07 |
| Sample | 114064 participants |
| intervention rate | 1536 deaths per 100000 person years |
| comparator rate | 1472 deaths per 100000 person years |
Harms: Not extracted.
Stated population/comparison only.
Extraction details
Access: Indexed primary abstract and disclosures
Funding: NIH
Conflicts: Author disclosed PAHO consulting.
Query: site:pubmed.ncbi.nlm.nih.gov ultra processed food mortality 2024 30 year cohort Fang
Certainty ungraded. Appraisal and qualified review pending.
Why might a diet heavy in these foods be a problem?
One important pathway is excess energy intake: many fast foods and snack formulations put substantial calories into convenient, highly palatable portions. WCRF’s assessment links dietary patterns rich in these energy-dense foods to weight gain. This concerns the pattern and food properties, not proof that every NOVA group-4 item has the same effect.[14]
What foods replace also matters. WCRF separately assesses wholegrains, vegetables, fruit and beans, including their fibre and nutrient contributions. A pattern that displaces these foods with energy-dense products can change the overall diet. This is a dietary explanation; it does not identify every unfamiliar additive as a toxin.[15]
The short feeding experiments below examine actual intake and texture. WCRF’s recommendation uses a different food-pattern definition from NOVA: it targets fast foods and processed foods rich in fat, refined starches or sugars. A processing category, a nutrient profile and a demonstrated clinical effect therefore remain different descriptions.[14]
What controlled feeding trials found
In Hall and colleagues’ 2019 inpatient crossover trial, 20 adults ate each of two menus for two weeks. They could eat freely. Intake averaged about 508 calories a day more on the ultra-processed menu, with about 0.9 kg gained versus 0.9 kg lost on the other menu. The offered menus were designed to match several nutrients and calories; the amounts actually eaten differed.[2]
The trial tested entire menus, not one additive or processing step. Faster eating and differences in the foods offered are possible contributors. Randomization supports an effect of those menus in that setting, but does not isolate one mechanism or establish identical effects for every product.[2]
In Dicken and colleagues’ 2025 crossover trial, 55 adults with overweight or obesity were randomized to two eight-week diets, both following the UK Eatwell Guide; 50 were included in the intention-to-treat sample. Weight fell on both diets: about 2.06% with minimally processed food and 1.05% with UPFs. The difference was 1.01 percentage points (95% confidence interval 0.14–1.87) favoring minimally processed food. Mild gastrointestinal adverse events were common with both.[3]
The comparison matters: a UPF menu meeting dietary guidance did not produce the same weight trajectory as the 2019 menu. These short trials do not establish long-term disease outcomes or a personal weight-loss prediction. The 2025 trial was published after SACN’s rapid assessment.[3][2]
What long-term studies add
Lane and colleagues’ 2024 umbrella review combined findings from 45 pooled analyses of observational research. Higher UPF exposure was associated with several adverse outcomes. The authors rated the underlying evidence as moderate for four analyses, low for 22, and very low for 19. Strong statistical associations and high certainty are not interchangeable: this review does not demonstrate that ultra-processing causes every associated disease.[6]
A 2024 EPIC cohort analysis followed 311,892 adults for an average 10.9 years and identified 14,236 new cases of type 2 diabetes. It reported a higher incidence with greater UPF consumption, but results differed between UPF subgroups. This adds long-term observations; it does not randomly assign diets or prove that every food in group 4 has the same effect.[8]
Several reviews can include the same underlying cohorts, so their reference counts are not counts of independent replications. Observational studies can also differ in how they record diet, classify foods, adjust for other behaviors, or choose comparisons. Residual confounding means that relevant differences between people may remain after statistical adjustment.[6]
A processing group does not fix eating rate or energy intake
In a trial published online in November 2025 and in an April 2026 issue, 41 adults completed two 14-day UPF menus selected for different textures and eating rates. The menus matched offered energy, palatability, variety and non-beverage energy density. Intake was 369 kcal/day lower on the slower-eating menu, with a 95% confidence interval of 221–517.[12]
Body weight did not differ clearly between these menus. The comparison shows that texture-related intake differences can exist within the same processing category. It does not isolate every manufacturing step, prove a long-term disease effect or supply a calorie reduction everyone can reproduce by counting chews.[12]
Grams per minute and calories per gram describe different properties
Eating rate describes how quickly food is consumed; energy density describes energy per weight of food. A 2025 crossover experiment gave 69 adults five lunchtime meals varying these properties. The slow, lower-density sandwich produced substantially lower energy intake than the fast, higher-density sandwich.[13]
This was a meal experiment in young, generally healthy adults, not a lifetime dietary comparison. Texture, energy density and processing labels should therefore be interpreted together. The investigators disclosed food-industry relationships; those disclosures belong alongside design and replication when assessing the findings.[13]
What researchers disagree about—and what they share
Proponents argue that processing-based categories add useful information about the food system, manufacture and dietary displacement. In a 2026 policy analysis, Monteiro and colleagues present NOVA-based policies as complementary to nutrient-based measures. That is an argument about how to use accumulating evidence in policy; it does not establish that every group 4 product has the same causal effect.[17]
A 2026 critical analysis argues that manufacturing categories can obscure nutritional quality, time scarcity, income and food access. It challenges causal interpretations of a formulation label and favors policies emphasizing nutrients and equity. This critique does not negate the randomized menu findings; those trials and the broader category-wide causal claim ask different questions.[18]
An empirical 2026 comparison classified 4,605 food items using four systems and analyzed 4,392 NHANES adults. Systems assigned different proportions to highly processed categories, yet several nutrient and cardiometabolic associations pointed in similar directions. This cross-sectional comparison suggests both shared signals and classification differences; it cannot resolve causality or show which individual product causes disease.[19]
A percentage depends on how foods were classified
In a 2023 analysis of 958 Chilean preschoolers’ dietary records from 2016, researchers compared three ways of applying NOVA. Estimated calorie share from UPF was about 47% using food descriptions, 52% after checking ingredient lists for selected UPF features and 58% with a broader search for additives. Some additive functions were unknown, and ingredients were unavailable for some foods. This is a measurement comparison, not a trial of health effects. A percentage therefore needs its coding method, food information, population and year.[26]
Why uncertainty remains
SACN’s 2025 assessment considered the associations concerning while finding processing-specific effects difficult to separate from nutritional and other factors.[5]
Classification uncertainty and uncertainty about harm are distinct. A product can clearly meet a NOVA definition while its specific long-term effect remains uncertain. Conversely, an ambiguous classification does not mean its nutrient content or other health evidence is unknowable.[7]
References
- Ultra-processed foods: what they are and how to identify them
Classification commentary
Monteiro CA and colleagues · Content/publication date: 2019
Accessed: 2026-10-06 UTC · Selected publisher full-text sections
Cited sections: Defining ultra-processed foods; Non-ultra-processed food groups; Identifying ultra-processed foodsClassification commentary.
- Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake
Randomized crossover trial
Hall KD and colleagues · Content/publication date: 2019
Accessed: 2026-10-06 UTC · Indexed abstract and figure captions
Cited sections: Abstract; Figure 2: food intake and eating rateLinked errata; independent full trial appraisal pending.
- Ultraprocessed or minimally processed diets following healthy dietary guidelines on weight and cardiometabolic health: a randomized, crossover trial
Randomized crossover trial
Dicken SJ and colleagues · Content/publication date: 2025
Accessed: 2026-10-06 UTC · Abstract and publication record
Cited sections: Abstract: participants, intention-to-treat sample, percent weight change, adverse eventsPublished online 4 August 2025. Outcomes are from the abstract; no long-term disease conclusion is inferred.
- 2017/2018 POF: fresh food and culinary preparations prevail in the national diet
Official classification explanation
Brazilian Institute of Geography and Statistics (IBGE) · Content/publication date: 2020
Accessed: 2026-10-06 UTC · Indexed official-page text; direct retrieval was blocked
Cited sections: Paragraph describing the four NOVA groups and examplesUsed for the category table, not a new clinical-effect claim.
- Processed foods and health: SACN’s rapid evidence update summary
Official evidence assessment
UK Scientific Advisory Committee on Nutrition · Content/publication date: 2025
Accessed: 2026-10-06 UTC · Official summary text
Cited sections: Results; Limitations; Discussion; ConclusionsPrecedes August 2025 trial; rapid update without formal study-quality assessment.
- Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses
Umbrella review of observational studies
Lane MM and colleagues · Content/publication date: 2024
Accessed: 2026-10-06 UTC · Indexed primary-article abstract and text; direct retrieval unavailable
Cited sections: Methods; Results: GRADE certainty; Discussion: bias and confoundingThe reported certainty grades belong to this publication; the Atlas has not independently regraded the evidence.
- Ultra-processed foods: how functional is the NOVA system?
Classification reliability study
Braesco V and colleagues · Content/publication date: 2022
Accessed: 2026-10-06 UTC · Abstract and disclosure statement
Cited sections: Abstract: methods, agreement, conclusionsExamines classification agreement, not disease causation. Some authors disclosed food-industry relationships.
- Food consumption by degree of food processing and risk of type 2 diabetes mellitus: a prospective cohort analysis of EPIC
Prospective cohort study
Dicken SJ and colleagues · Content/publication date: 2024
Accessed: 2026-10-06 UTC · Indexed primary-article summary
Cited sections: Summary: methods, findings, subgroup heterogeneityDiet was assessed with questionnaires. The study is observational; modeled substitutions are not randomized food swaps.
- Food Additives and GRAS Ingredients — Information for Consumers
Agency explanation
US Food and Drug Administration · Content/publication date: not stated
Accessed: 2026-10-06 UTC · Official page text
Cited sections: Functions of ingredients; Safety standard and oversightUsed to distinguish ingredient functions and safety assessment from dietary-pattern research.
- High-pressure processing: food safety without compromising quality
Agency assessment summary
European Food Safety Authority · Content/publication date: 2022-03-08
Accessed: 2026-10-06 UTC · Official summary of an EFSA scientific opinion
Cited sections: Description and applications of high-pressure processingA food-safety processing example, not evidence that every processed food is nutritionally equivalent.
- Are all ultra-processed foods bad? A critical review of the NOVA classification system
Critical narrative review
Louie JCY · Content/publication date: 2025-08-04
Accessed: 2026-10-06 UTC · Selected publisher full-text sections
Cited sections: Methodological challenges in research: bread categories and food-frequency questionnairesUsed for concrete classification examples and dietary-measurement limitations. This narrative review is not a new randomized trial.
- Eating rate has sustained effects on energy intake from ultraprocessed diets: a 2-week ad libitum dietary randomized controlled crossover trial
Randomized crossover trial
Forde CG and colleagues · Content/publication date: 2026
Accessed: 2026-10-08 UTC · Original indexed abstract and selected original journal result passages inspected; direct PMC access showed a verification wall
Cited sections: Abstract: two 14-day UPF menus, eating rate, intake and body weightOnline 26 November 2025; April 2026 issue. Selected source, not independent full appraisal. No verification wall was bypassed.
- The Effect of Energy Density and Eating Rate on Ad Libitum Energy Intake in Healthy Adults—A Randomized Controlled Study
Randomized crossover meal experiment
Lasschuijt MP and colleagues · Content/publication date: 2025
Accessed: 2026-10-08 UTC · Original indexed abstract and relevant passages inspected; direct retrieval unavailable
Cited sections: Abstract, five lunch conditions and conflict-of-interest disclosureFood-industry speaking/travel relationships disclosed. An acute meal experiment, not a clinical disease-outcome trial.
- Limit fast foods: recommendation evidence
Professional evidence assessment
World Cancer Research Fund · Content/publication date: not stated
Accessed: 2026-10-10 UTC · Relevant original organization assessment passages inspected; no independent synthesis of every underlying study
Cited sections: Original energy-intake, food-pattern, portion and weight-gain explanation; definition differs from NOVA - Wholegrains, vegetables, fruit and beans: recommendation evidence
Professional evidence assessment
World Cancer Research Fund · Content/publication date: not stated
Accessed: 2026-10-10 UTC · Relevant original assessment passages inspected; selected dietary-pattern explanation, not individual product appraisal
Cited sections: Original dietary fibre/nutrient and food-replacement interpretation - Definitions of ultra-processed foods beyond NOVA: a systematic review and evaluation
Systematic classification review
Medin and colleagues · Content/publication date: 2025
Accessed: 2026-10-13 UTC · Selected original source passages inspected
Cited sections: Abstract: objectives, methods, identified classification systems and comparison limitationsTargeted educational retrieval; no independent full-text appraisal.
- What Do Nova and the Ultraprocessed Food Concept Offer to Policymakers?
Policy analysis
Monteiro and colleagues · Content/publication date: 2026
Accessed: 2026-10-13 UTC · Original abstract inspected; policy argument, not a randomized clinical experiment
Cited sections: Abstract: proposed policy applications and relationship to nutrient-based policiesTargeted educational retrieval; no independent full-text appraisal.
- The industrial formulation fallacy: a critical appraisal of the Nova classification system
Critical narrative analysis
Jimmy Chun Yu Louie · Content/publication date: 2026
Accessed: 2026-10-13 UTC · Original abstract inspected; author interpretation, not a definitive causal assessment
Cited sections: Abstract: causal critique, structural determinants and policy interpretationTargeted educational retrieval; no independent full-text appraisal.
- Comparison of food processing classification frameworks in NHANES2017–2018
Cross sectional classification comparison
Du and colleagues · Content/publication date: 2026
Accessed: 2026-10-13 UTC · Original abstract and selected reported methods/results inspected
Cited sections: Abstract: NHANES sample, food coding, agreement and cross-sectional associationsTargeted educational retrieval; no independent full-text appraisal.
- UPF in a small feeding trial
Randomized open label crossover trial
Hamano S and colleagues · Content/publication date: 2024
Accessed: 2026-10-14 UTC · Indexed primary abstract and publisher methods/results
Cited sections: Abstract; Methods; Results: body weightExtraction; independent review pending.
- UPF and mortality in US cohorts
Prospective cohort study
Fang Z and colleagues · Content/publication date: 2024
Accessed: 2026-10-14 UTC · Indexed primary abstract and disclosures
Cited sections: Abstract: Methods/Results; Conflict of interestExtraction; independent review pending.
- Ultra-processed foods, diet quality, and health using the NOVA classification system
Agency commissioned classification report
Monteiro and colleagues / FAO · Content/publication date: 2019
Accessed: 2026-10-15 UTC · Selected report text, author-uploaded copy inspected; official repository retrieval failed
Cited sections: Section 2, pages 6–12: groups and ingredient examplesAuthor-uploaded copy and FAO repository identify the same report; one publication budget.
- Dietary Guidelines for the Brazilian Population, second edition
Government dietary guidance
Ministry of Health of Brazil · Content/publication date: 2014
Accessed: 2026-10-15 UTC · Selected original guideline text inspected through a hosted copy; original ministry URL timed out
Cited sections: Chapter 2; pages 79–80 plain versus formulated yogurtsOriginal ministry publication; hosted copy is the same publication.
- Ultra-Processed Foods
Agency explanation
US Food and Drug Administration · Content/publication date: not stated
Accessed: 2026-10-15 UTC · Official page text, selected relevant sections
Cited sections: Accelerating Research; Defining UPFs; historical request for information - Food Labeling Guide, March 2025
Government labeling guidance
Michigan Department of Agriculture and Rural Development · Content/publication date: 2025
Accessed: 2026-10-15 UTC · Original six-page PDF, selected relevant sections
Cited sections: Pages 2–3: ingredient lists and Nutrition Facts - Impact of the use of food ingredients and additives on the estimation of ultra-processed foods and beverages
Original cross sectional methods study
Ricardo and colleagues, Frontiers in Nutrition · Content/publication date: 2023
Accessed: 2026-10-20 UTC · Original publisher full text: relevant methods, results, limitations, funding and conflict statements read.
Cited sections: Methods2.1–2.4, Results3 and Discussion4: dietary sample, three coding methods, estimates and limitationsMeasurement comparison, not a health-outcome trial.
Scope & review
- Population
- Definitions for general readers; adult feeding trials and adult cohorts
- Exposure or intervention
- Ultra-processed versus minimally processed dietary patterns
- Comparison
- Controlled dietary patterns; food-category definitions
- Outcomes
- Food classification; energy intake; body-weight change; selected observational disease associations
- Timeframe
- Two- and eight-week feeding periods; years of observational follow-up
- Jurisdiction
- International research; Brazilian NOVA examples; UK assessment; US and EU ingredient/processing explanations
Selected-source educational explanation. Edition 3.18 adds scoped explanations and records source-specific access or recheck dates. Older references retain their own dates. No exhaustive search, independent systematic synthesis, formal Atlas grade or qualified clinical review has been completed.
Sources checked on their individual access dates. Selected sources were checked; ongoing surveillance and qualified review have separate statuses on the coverage page.
Revision history
- 2026-10-05 UTC: Initial selected-source overview.
- 2026-10-06 UTC: Scope and source locators added; repetitive prose condensed. Original source-check date retained; no new claim review.
- 2026-10-06 UTC: Unverifiable generated search-query labels replaced with retained reference URLs; source-check date and clinical prose preserved.
- 2026-10-06 UTC: Expanded the definition, four NOVA groups, processing purposes, label limitations, nutritional distinctions, trial detail, observational evidence, and uncertainty; increased references from 3 to 11 and retained existing question links. Added bread and cereal formulation comparisons.
- 2026-10-06 UTC: Added related reading and made remaining editorial scope explicit in edition 3.1; source-access dates retained.
- 2026-10-08 UTC: Deepened specific reader explanations and added or rechecked the recorded sources; other earlier content retained.
- 2026-10-10 UTC: Edition 3.41: added an early plain-language explanation of energy intake and dietary replacement, two reader questions and two organization assessment references. Existing passages, source dates and anchors retained.
- 2026-10-13 UTC: Added comparison of processing classification systems and placed definitions/types before evidence; selected2025 classification review inspected. Existing source access dates retained.
- 2026-10-15 UTC: Added the plain-English series explanation, practical questions and selected primary sources; independent review remains pending.
- 2026-10-20 UTC: Added an original classification-method study and a question explaining why UPF intake estimates can differ. No new health-outcome conclusion or qualified review claimed.
What this article still leaves open
- How reliably do classifiers agree on mixed dishes and changing branded products?
- How much do formulation, food structure, eating rate and dietary replacement explain observed outcomes?
- Which long-term randomized comparisons can separate processing from nutritional composition?
- Further controlled texture/energy-density comparisons, independent replication, long-term outcomes and review-overlap reconciliation remain necessary.
Suggest a correction
Submit a content correction to the editorial queue, or download a local note. Leave out personal medical details. Submitted text is stored privately with a browser receipt session and is visible to this browser and the editor. Clearing cookies removes access to your submission history. This is not an account or a medical service.
Your submitted corrections for this article
More in the plain-English series
Keep making sense of it.
Explore Food & nutrition explanations See every explanationExplore the topic map
Browse the health domains connected to this article. The map also identifies topics whose articles remain unwritten.