What counts as an ultra-processed food?

Under the NOVA classification system, ultra-processed foods are industrial formulations typically made mostly or entirely from substances extracted or refined from foods (starches, oils, sugars, isolated proteins) plus additives rarely used in home cooking, such as emulsifiers, flavor compounds, and colorings. Common examples include packaged snack cakes, sugary breakfast cereals, reconstituted meat products, and many diet sodas. The category is defined by formulation and processing method, not by calories, fat, or sugar content alone.

Is all processed food bad for you?

No. NOVA distinguishes several tiers, and most "processed" foods — canned beans, cheese, bread made with a few recognizable ingredients, cured meats — sit in a less-processed tier than ultra-processed foods and are not treated the same way in the research. Processing itself is not the concern; the category researchers are most focused on is the specific industrial formulation style NOVA labels ultra-processed.

Does eating ultra-processed food cause weight gain?

One small, tightly controlled feeding trial found people ate more calories and gained weight on an ultra-processed diet compared with a minimally processed diet matched for calories, sugar, fat, fiber, and macronutrients — suggesting a possible causal mechanism. Most of the broader evidence linking ultra-processed intake to weight gain and chronic disease, however, is observational, meaning it can show an association but cannot on its own prove the food processing itself is the cause rather than other factors that track along with it.

Oct 12, 2026· By reConnect

Ultra-Processed Foods: What the NOVA Classification Actually Shows

NOVA sorts food by how processed it is, not by nutrients. Here is what the ultra-processed research actually supports, and where the science is still catching up.

#diet-quality #food-science #nova-classification #nutrition-science #ultra-processed-foods #whole-foods

Walk into a nutrition conversation in the last few years and “ultra-processed food” comes up almost as often as “sugar” or “saturated fat” used to. The term comes from a specific research framework called NOVA, and it has become shorthand for “food you should avoid.” The framework is genuinely useful, but the shorthand oversimplifies it. Here is what NOVA actually measures, what the research linking ultra-processed food to health outcomes actually shows, and where the confident-sounding claims outrun the evidence.

What NOVA actually sorts

NOVA is a food classification system developed by Brazilian researchers that groups foods into four categories based on the extent and purpose of industrial processing, not on their nutrient content:

Unprocessed or minimally processed foods — fresh, dried, or frozen produce, plain meat, eggs, milk.
Processed culinary ingredients — oils, butter, sugar, salt, used to prepare food at home.
Processed foods — a combination of the first two groups, made largely recognizable, such as canned vegetables, cheese, or freshly baked bread.
Ultra-processed foods — industrial formulations built mostly from substances extracted or synthesized from foods (isolated starches, protein isolates, refined oils, sugars) plus additives — emulsifiers, flavorings, colorings, texturizers — that rarely appear in a home kitchen.

The defining feature is the formulation and manufacturing process, not calories, fat, sugar, or any single nutrient. A packaged food can be low in sugar and still classified ultra-processed if its ingredient list is built the way NOVA describes; a homemade dessert loaded with sugar and butter is not ultra-processed under this system at all, because it uses culinary ingredients rather than industrial ones. That distinction trips people up constantly, and it is worth sitting with before assuming NOVA is just a fancier way of saying “junk food.”

The evidence that looks genuinely strong

The most-cited piece of direct evidence is a 2019 NIH inpatient feeding trial led by Kevin Hall, in which participants lived on a metabolic ward and were given, in a randomized crossover design, two weeks each of an ultra-processed diet and a minimally processed diet, carefully matched for calories, sugar, fat, fiber, and macronutrients on paper. Participants spontaneously ate roughly 500 more calories a day on the ultra-processed diet and gained about 0.9 kg (2 lb) over the two weeks, then lost a similar amount on the minimally processed one (Hall et al., Cell Metabolism, 2019). Because the study was randomized and controlled — not just observed in the wild — it is one of the stronger pieces of evidence that something about the ultra-processed formulation itself, beyond the nutrition label, drives people to eat more. It was also small (20 adults) and short (two weeks per phase), and the meals were matched for nutrients as offered, not as eaten, so it points to a plausible mechanism rather than settling the question for all diets, populations, and longer time frames.

Proposed mechanisms include faster eating rate (participants in the Hall trial did eat the ultra-processed meals faster), higher energy density packed into less satiating volume, and flavor-and-texture combinations engineered to be highly palatable. These are hypotheses still being tested, not established explanations — but each could plausibly push people to eat more calories without feeling as full.

The evidence that is weaker than the headlines suggest

Beyond that one feeding trial, most of the research connecting ultra-processed food intake to obesity, type 2 diabetes, cardiovascular disease, and even mortality comes from large observational cohort studies. These studies are genuinely useful for spotting patterns across populations, and the direction of the pattern here — higher ultra-processed intake associated with worse average health outcomes — shows up fairly consistently across different countries and cohorts. The strength of that evidence varies a lot by outcome, though: a 2024 umbrella review in The BMJ graded the links with outcomes such as cardiovascular death, type 2 diabetes, and common mental health problems as among the more robust, while many of the other associations it examined were graded only suggestive, weak, or not supported, and the individual studies differ considerably in how they measure intake. But observational data cannot, by itself, prove the processing is the cause rather than a marker for something else. People who eat a lot of ultra-processed food also tend to eat less fiber, less produce, and more added sugar and refined starch on average — any of which could independently explain some of the association, alongside non-dietary factors like income, health literacy, or access to fresh food.

Researchers in this space are generally careful about this distinction; a lot of popular coverage is not. “Ultra-processed food causes disease X” is a stronger claim than the current evidence base, taken as a whole, actually supports — “ultra-processed food is consistently associated with worse outcomes, and there’s a plausible causal mechanism from at least one controlled trial” is closer to accurate.

A practical way to use the category

Given where the evidence actually stands, a reasonable approach is to use NOVA as a rough sorting tool rather than a strict rulebook:

Build meals around less-processed foods most of the time — produce, whole grains, plain proteins, dairy, and legumes.
Don’t panic over an ingredient list with an emulsifier or two. A loaf of bread with one added preservative is a different thing, nutritionally and by volume in most diets, than a diet built primarily around packaged snack foods and sugary drinks.
Pay attention to volume and pattern, not single products. The research concern is about diets where ultra-processed foods make up the bulk of calories, not the occasional packaged item in an otherwise varied diet.
Watch for eating speed and fullness cues if you do eat more processed convenience foods regularly — the proposed mechanism from the feeding trial is about how fast and how much people eat, which is something you can pay attention to directly, regardless of how the science eventually settles.

The honest takeaway

NOVA is a genuinely useful lens for thinking about how much of a diet comes from foods built the way a kitchen would build them versus foods built the way a factory formulates them. The controlled feeding evidence suggests that distinction may matter beyond the numbers on a nutrition label. But most of what connects ultra-processed eating to long-term disease risk is still observational, and treating every packaged food with an unfamiliar ingredient as equally dangerous overstates what one small trial and a body of correlational research can support. The evidence justifies leaning toward whole and minimally processed foods as a general default — it does not yet justify sorting every food into “safe” or “poison” based on its NOVA category alone.

References

Hall, K. D., 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. PubMed
Lane et al. (2024). Ultra-processed food exposure and adverse health outcomes: umbrella review. The BMJ. BMJ