What is ultra-processed food?

Ultra-processed food, under the NOVA classification by Carlos Monteiro at the University of Sao Paulo, is food produced from industrial extracts of food components combined with additives not found in any home kitchen: protein hydrolysates, modified starches, emulsifiers, color additives, and flavor compounds. Over 70% of the US food supply falls into this category.

What did the NIH study on ultra-processed food find?

Kevin Hall at the NIH ran a randomized crossover trial with 20 adults. Both an ultra-processed diet and a whole-food diet were matched for the same macronutrient levels: fat, sugar, and fiber. On the ultra-processed diet, participants consumed an average of 508 more calories per day and gained weight. On the whole food diet, they consumed less and lost weight.

How do I know if food is ultra-processed?

Look at the ingredient list. If any ingredient is not a recognizable whole food or standard kitchen ingredient, the food is likely ultra-processed. Specific signals include protein hydrolysates, modified starch, maltodextrin, high-fructose corn syrup, color additives by number, generic natural or artificial flavors, and emulsifiers by chemical name. The practical test: could you make this at home from real ingredients?

Does ultra-processed food cause weight gain?

The only randomized controlled trial found that participants gained approximately 2 pounds over two weeks on a matched ultra-processed diet versus losing approximately 2 pounds on a whole food diet. Macronutrients were matched, so the difference was not attributable to greater fat or sugar. A 2024 umbrella review of 45 meta-analyses found highly suggestive evidence for an association between ultra-processed food and obesity.

Why do dietary guidelines not mention ultra-processed food?

The USDA dietary guidelines have never used the term ultra-processed. During the 2020 guideline cycle, the International Food Information Council and the Grocery Manufacturers Association opposed any guidance framing processing level as a health determinant. The USDA also administers crop subsidies for corn and soybeans, the primary raw materials of the ultra-processed food supply.

Ultra-Processed Food Is Not Food. An NIH Randomized Controlled Trial Found Participants Consumed 508 More Calories Per Day Without Knowing Why.

Ultra-processed food is not food with some processing done to it. It is a distinct industrial category: formulations produced from substances extracted from food, combined with additives such as flavors, colors, emulsifiers, and sweeteners, in ways that cannot be replicated in a home kitchen. Over 70% of the US food supply falls into this category. Most Americans eat it at every meal without a working definition of what it is.

Kevin Hall at the National Institute of Diabetes and Digestive and Kidney Diseases ran the only randomized controlled trial ever conducted on ultra-processed food. His participants consumed 508 more calories per day on the ultra-processed diet than on the whole food diet. Both diets were matched for total calories offered, macronutrients, sugar, fat, and fiber. The participants were not told to overeat. Their bodies simply stopped registering that they had eaten enough.

What Ultra-Processed Food Actually Is

Carlos Monteiro, a nutritional epidemiologist at the University of Sao Paulo, developed the NOVA food classification in 2009. NOVA does not classify food by its nutritional content. It classifies food by what was done to produce it. This distinction is the entire point.

NOVA Group 1: Unprocessed or minimally processed foods. Whole vegetables, fruit, meat, eggs, legumes, grains, and nuts. Foods essentially as they came from nature. Freezing, drying, fermenting, and pasteurizing keep food in Group 1. The food is still recognizably its source ingredient.

NOVA Group 2: Processed culinary ingredients. Oils, butter, flour, sugar, and salt. Substances extracted from Group 1 foods are used in home cooking. Not eaten alone. Used to prepare real food from real ingredients.

NOVA Group 3: Processed foods. Food made by combining Group 1 and Group 2 in ways that preserve or alter texture, flavor, or shelf life. Canned vegetables with salt. Cured meats. Real cheese. Processed, but not ultra-processed.

NOVA Group 4: Ultra-processed food and drink products. Industrial formulations made from substances extracted from foods, with or without the food itself also present. These products contain ingredients not found in any home kitchen: protein hydrolysates, modified starches, hydrogenated oils, interesterified fats, maltodextrin, high-fructose corn syrup, glucose syrup, carboxymethylcellulose, polysorbate 80, and a full range of flavor compounds, color additives, and emulsifiers. These are not food. They are industrial inputs assembled into something that resembles food.

The practical test Monteiro recommends: could you make this at home with ingredients from a grocery store? A loaf of bread made with flour, water, yeast, and salt: yes. A commercial packaged bread with 28 ingredients, including calcium propionate, azodicarbonamide, and diacetyl tartaric acid esters of mono and diglycerides: no. If the product cannot be replicated in a home kitchen from real food, it qualifies as Group 4.

(Evidence level: 5 for the NOVA classification framework as a published, peer-reviewed taxonomy of food by production method. The classification is a descriptive system, not itself a health claim.)

The NIH Trial: 508 Extra Calories and Why Matching Macronutrients Changed Everything

Kevin Hall's 2019 study, published in Cell Metabolism, was a randomized crossover trial with 20 adult participants who stayed at a clinical research facility for four weeks. Each participant spent two consecutive two-week periods eating either an ultra-processed diet or a whole food diet, in randomized order. Both diets were matched as closely as possible for total calories offered, total fat, total carbohydrate, total protein, total sugar, and total fiber. Participants were instructed to eat as much or as little as they wanted at every meal.

On the ultra-processed diet, participants consumed an average of 508 more calories per day than on the whole food diet. They ate faster. They finished more of what was served. Over two weeks, they gained approximately 2 pounds. On the whole food diet, they ate less spontaneously and lost approximately 2 pounds.

The macronutrient matching is why this study matters and why the food industry response has focused on attacking its methodology rather than its findings. The most common industry counterargument to ultra-processed food research is that associations with overconsumption and weight gain are explained by differences in fat, sugar, or caloric density between UPF and whole foods. Hall's design removed that explanation by matching those variables. The excess calorie consumption could not be attributed to more sugar or more fat being offered. Something about the food's ultra-processed character drove the difference.

Hall's team found that participants on the ultra-processed diet ate faster, which compressed the time available for satiety hormones to signal fullness. The GLP-1 satiety hormone response was lower on the ultra-processed diet. PYY, a second satiety hormone, was also lower. The ultra-processed diet was associated with measurably impaired satiety signaling compared with the whole-food diet that provided the same macronutrient profile.

The International Food Information Council, funded by major food and beverage manufacturers, published commentary arguing that the Hall study's sample size was too small to draw conclusions. The sample size is a real limitation. It is also the reason the study required participants to live in a clinical facility for four weeks, which allowed Hall to control the diet precisely enough to match macronutrient intake and measure actual intake. Larger studies at this level of dietary control are not logistically achievable. The epidemiological evidence from millions of participants provides the scale that the controlled trial cannot.

(Evidence level: 4. Single randomized controlled trial with 20 participants in a crossover design. Strong methodology for dietary research. Findings have not yet been replicated in a larger RCT. Epidemiological evidence from multiple large independent cohorts is consistent with the trial findings.)

What Ultra-Processing Does to Food to Make This Happen

The Hall study showed the effect. The mechanism question is partially answered.

Speed of digestion: Ultra-processed foods are typically pre-digested during industrial processing. The food matrix, the physical structure of whole food that slows absorption, has been disrupted. Ultra-processed bread is softer, denser, and more rapidly absorbed than intact grain bread, even when fiber content is matched on a label. The disrupted food matrix means glucose enters the bloodstream faster, insulin responds more aggressively, and blood sugar falls more steeply. Satiety hormone release depends partly on the rate of digestion and the presence of food in the gut, not only on total calories consumed.

Hyperpalatability engineering: Ultra-processed products are formulated to hit specific targets of sweetness, saltiness, and fat content that stimulate reward pathways in the brain more aggressively than whole foods. The food industry employs sensory scientists at companies such as International Flavors and Fragrances and Givaudan, whose documented purpose is to identify formulation parameters that maximize consumption. Industry documents from internal research programs have used the phrase "bliss point" to describe the optimal convergence of sugar, fat, and salt that maximizes palatability. The stated function of these teams is not to satisfy hunger. It is to maximize the desire to keep eating.

Caloric density and eating speed: Ultra-processed foods are calorie-dense, tend to require less chewing, and are often served in portions that make the quantity ambiguous. Whole foods take longer to chew, slow the rate of eating, and provide physical volume that activates stomach stretch receptors, contributing to satiety. Ultra-processed food bypasses most of these mechanisms. Hall's participants on the ultra-processed diet ate faster and consumed more volume before satiety signaled adequacy.

(Evidence level: 4 for speed of digestion effects on satiety timing. Evidence level: 4 for hyperpalatability engineering as industry practice, documented in corporate records and published research. Evidence level: 3 for the specific mechanistic pathway connecting these to the 508-calorie excess specifically.)

70% of the US Food Supply and the Disease Data

70% of the US food supply qualifies as ultra-processed under the NOVA classification. Only about 4% is minimally processed. This is not a marginal segment of the American diet. For most Americans, ultra-processed food is the default, and whole food is the exception.

The epidemiological evidence associating ultra-processed food with disease outcomes is now substantial and consistent across multiple research groups with no financial relationship to the food industry.

The NutriNet-Sante cohort in France followed 105,000 adults over multiple years. A 2018 BMJ paper by Fiolet, Srour, and colleagues found that a 10% increase in ultra-processed food consumption was associated with a 12% higher risk of overall cancer. A 2019 BMJ paper by Srour and colleagues, using the same cohort, found that higher UPF intake was associated with higher cardiovascular disease risk and cardiovascular mortality.

The strongest single evidence summary is the 2024 BMJ umbrella review by Lane and colleagues, which analyzed 45 meta-analyses covering approximately 10 million study participants. Its findings by evidence tier:

Convincing evidence for associations with cardiovascular disease mortality, type 2 diabetes, anxiety, and depression. Highly suggestive evidence for all-cause mortality, sleep disorders, and obesity.

The designation "convincing evidence" in an umbrella review carries a specific methodological meaning: multiple meta-analyses showing the same direction of effect, high statistical certainty across the body of evidence, and established biological plausibility. This is not the same as suggestive or preliminary evidence. It is the strongest evidence tier the framework recognizes.

The disease associations follow a consistent pattern: the more ultra-processed food in the diet, the higher the risk. The displacement effect that Moubarac and colleagues at the University of Montreal documented in 2017 adds context: as ultra-processed foods occupy more of the diet, whole foods that provide vitamins, minerals, and fiber occupy less of it. The micronutrient depletion effect compounds the direct health effects of ultra-processed foods themselves.

(Evidence level: 4 for disease associations. Multiple large independent epidemiological cohorts with consistent findings. A 2024 umbrella review of 45 meta-analyses constitutes the strongest available evidence synthesis outside of a large randomized trial. Causation established by the Hall RCT for the caloric intake mechanism; direct disease causation requires further controlled evidence.)

Why the USDA Dietary Guidelines Have Never Said Ultra-Processed

The phrase ultra-processed food does not appear in the 2020-2025 Dietary Guidelines for Americans. It has never appeared in any edition of the federal dietary guidelines. The guidance that shapes school lunch programs, SNAP food assistance, military nutrition, and public health messaging has not identified ultra-processing as a category to reduce.

The Dietary Guidelines Advisory Committee accepts public comment. During the 2020 cycle, the International Food Information Council, funded by major food and beverage manufacturers, submitted comments opposing any guidance that would frame processing level as a health determinant. The Grocery Manufacturers Association submitted comments in the same direction. Their stated argument is that the NOVA classification is too broad and provides an unreliable guide to nutritional quality. The same industry that produces the ultra-processed food supply determines whether its processing method should appear in federal dietary guidance.

The institutional conflict is structural. The USDA issues dietary guidelines and simultaneously administers the crop subsidy programs that support corn, soybean, and wheat production. High-fructose corn syrup, soybean oil, and refined wheat flour are primary inputs for the ultra-processed food supply. The institution advising Americans to eat better is the same institution that financially supports the raw material supply chain for the food that most directly contradicts that advice.

Follow the Funding

Kevin Hall's NIH trial was funded by the National Institutes of Health intramural research program. No food industry financial relationship. Carlos Monteiro's NOVA research at the University of São Paulo has been funded by the Brazilian Ministry of Health and international research foundations. The NutriNet-Sante cohort is funded by the French National Institute of Health and Medical Research, the French National Cancer Institute, and the French Ministry of Health. The Lane 2024 umbrella review was funded by Deakin University and Australian government health agencies.

The research establishing that ultra-processed foods drive overconsumption and disease came from government agencies and public universities with no financial ties to the food industry.

The International Food Information Council describes itself as a nonprofit that communicates science-based information on food, nutrition, and food safety. Its funding comes from food and beverage manufacturers whose revenue depends on the ultra-processed products the science most consistently indicts. IFIC's published positions on the NOVA classification have consistently argued against its adoption in dietary guidance. IFIC does not disclose the complete list of its corporate funders.

Nestlé, which produces thousands of ultra-processed products and is one of the world's largest food companies, has published its own nutritional research framework that categorizes foods by nutrient content rather than processing level. A framework centered on nutrients rather than processing avoids the implication of the Hall study: that matching nutrients does not equalize health outcomes between ultra-processed and whole food diets.

The research defending continued consumption of ultra-processed food has come from researchers and organizations with financial relationships to the industry that produces it. The research showing its health effects has come from public institutions without those relationships.

What Is Proven, Plausible, and Unknown

Proven (Level 4-5 evidence): Over 70% of the US food supply qualifies as ultra-processed under the NOVA classification framework (Level 4. NOVA database analysis). An ultra-processed diet led to 508 more calories per day than a matched whole-food diet in a randomized crossover trial (Level 4; Hall et al., Cell Metabolism, 2019). Satiety hormone responses to GLP-1 and PYY were lower on an ultra-processed diet compared to a matched whole-food diet (Level 3. Single study finding, requires replication). A 2024 umbrella review of 45 meta-analyses found convincing evidence for associations with cardiovascular mortality, type 2 diabetes, anxiety, and depression (Level 4. Lane et al., BMJ, 2024). Ultra-processed food is formulated using documented hyperpalatability engineering practices at major flavor companies (Level 4. Documented industry practice).

Plausible: mechanism documented, specific contribution not yet quantified (Level 2-3): The disrupted food matrix in ultra-processed foods accelerates glucose absorption compared to whole foods with matched carbohydrate content, which impairs the timing of satiety signaling (Level 3. Mechanistic support from food science literature, not yet directly confirmed in intervention trials). Reducing ultra-processed food intake produces measurable improvements in metabolic health markers, independent of total caloric intake (Level 3. Epidemiological data are consistent; RCTs not yet completed at this scale). The specific additives in ultra-processed foods, including emulsifiers and artificial sweeteners, contribute to health outcomes beyond the overconsumption effect alone (Level 2. Animal model data on emulsifiers are concerning; direct human evidence at typical food concentrations is limited).

Unknown (Level 1-2): The specific components within ultra-processed food that most drive the overconsumption effect, identified by Hall, and their individual magnitudes of contribution. Whether there is a threshold of ultra-processed food consumption below which the disease associations become negligible. The long-term effects of specifically ultra-processed food consumption on gut microbiome composition and whether those effects mediate the observed disease associations.

The Risk/Reward Verdict

ATH Verdict: Evidence Supports

The randomized controlled trial provides the experimental causal evidence. The epidemiological umbrella review of 45 meta-analyses provides evidence of population-scale disease associations. Together, they constitute a stronger evidence base than most dietary recommendations included in federal dietary guidance. Reducing ultra-processed foods and replacing them with whole foods does not require eliminating any essential nutrients, is not costly relative to the alternative, and is consistent with how humans ate for all of recorded history before the industrialization of the food supply. The evidence supports the reduction. The switch is achievable without eliminating entire food categories.

What To Do Today

Use the NOVA 30-second test on any packaged food. Read the ingredient list. If it contains any ingredient that is not a recognizable whole food or standard kitchen ingredient, it qualifies as Group 4. The specific triggers to look for: protein hydrolysates, modified starches, interesterified fats, maltodextrin, high-fructose corn syrup, glucose syrup, any color additive listed by name and number, any flavoring listed generically as "natural flavors" or "artificial flavors," emulsifiers by chemical name such as polysorbate 80, carrageenan, or carboxymethylcellulose.

Start with the highest-priority replacement targets:

Breakfast. Breakfast cereal, flavored yogurt, breakfast bars, packaged pastries, and flavored instant oatmeal are among the highest-concentration ultra-processed categories in the American diet. Replace with eggs, plain whole-milk yogurt with no added ingredients, plain rolled oats with real fruit, or leftover whole foods from the night before.

Beverages beyond water. Sweetened drinks of every kind, including drinks marketed as healthy, most protein shakes made from industrial protein isolates, and flavored sparkling water containing natural flavors. Replace with plain water, coffee, tea, or dairy milk.

Packaged snacks. Chips, crackers, cookies, granola bars, and most commercial trail mix contain Group 4 additives and were formulated to encourage overconsumption. Replace with whole nuts, fresh or dried fruit with no added sugar or oil, vegetables, or cheese made from four ingredients.

You do not need to eliminate all ultra-processed food. The Hall study showed that even partial replacement with whole food produces measurable differences in caloric intake and body weight over a two-week period. The reduction produces a benefit even when it is not an elimination.

Look for whole food products from verified producers in the ATH Food and Pantry collection. Every product has been evaluated against the NOVA Group 4 criteria before listing. No ultra-processed food products are listed on ATH.

Go Verify

Search PubMed for "Hall ultra-processed diets excess calorie intake Cell Metabolism 2019." Read the full paper. Pay attention to Figure 1, which shows the divergence in caloric intake between the two diet groups despite identical macronutrient offerings. Read the discussion section's account of why macronutrient matching did not equalize outcomes.
Search for "Lane ultra-processed food adverse health outcomes BMJ 2024." Read the Methods section on how the umbrella review defined "convincing evidence" versus "highly suggestive" before interpreting the findings. The distinction is technical and important.
Apply the NOVA test to five products in your kitchen right now. List the ingredients that cannot be purchased at a grocery store or produced in a home kitchen. Track the percentage of items in your current pantry that fail the test. This is more informative than any article about the topic.
Go to the International Food Information Council website at ific.org. Read their published position on the NOVA classification. Then read the funders and board members page. The funding relationships and the published positions are both public information.
Search the Federal Register for comments submitted to the 2020 Dietary Guidelines Advisory Committee by the Grocery Manufacturers Association and the International Food Information Council. The comments are public documents. Note specifically which guidance language was opposed and why.

Sources and Citations

Hall KD, Ayuketah A, Brychta R, et al. "Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake." Cell Metabolism. 2019;30(1):67-77.
Monteiro CA, Cannon G, Levy RB, et al. "Ultra-processed foods: what they are and how to identify them." Public Health Nutrition. 2019;22(5):936-941.
Lane MM, Gamage E, Du S, et al. "Ultra-Processed Food Exposure and Adverse Health Outcomes: Umbrella Review of Epidemiological Meta-Analyses." BMJ. 2024;384:e077933.
Fiolet T, Srour B, Sellem L, et al. "Consumption of ultra-processed foods and cancer risk: results from NutriNet-Sante prospective cohort." BMJ. 2018;360:k322.
Srour B, Fezeu LK, Kesse-Guyot E, et al. "Ultra-processed food intake and risk of cardiovascular disease: prospective cohort study." BMJ. 2019;365:l1451.
Moubarac JC, Batal M, Louzada ML, Martinez Steele E, Monteiro CA. "Consumption of ultra-processed foods predicts diet quality in Canada." Appetite. 2017;108:512-520.
Moss M. "Salt Sugar Fat: How the Food Giants Hooked Us." Random House. 2013.