Nearly all food is processed to some degree — cooking, freezing, canning, and drying are all forms of processing, and none of them are what current nutrition research is actually concerned about. The real distinction driving recent research isn’t “processed versus not,” it’s a specific category at the far end of that spectrum: ultra-processed food.
This distinction matters because “avoid processed food” as blanket advice is nearly impossible to follow and, as covered below, throws out some of the more practical and nutritious options available along with the ones actually worth a second look.
The NOVA classification system
NOVA is a widely used framework, developed by researchers in Brazil and since adopted in various forms by international health bodies, that sorts food into four groups by processing level rather than by traditional nutrient categories: unprocessed or minimally processed foods (fresh or frozen vegetables, plain meat, eggs), processed culinary ingredients (oils, butter, sugar — used to prepare other foods rather than eaten alone), processed foods (canned vegetables in brine, cheese, freshly made bread), and ultra-processed foods (packaged snacks, sugary drinks, reconstituted meat products). Most of the current research attention, and most of the health concern, concentrates specifically on that fourth category.
What actually defines “ultra-processed”
Ultra-processed foods are industrial formulations built from ingredients rarely found in a home kitchen — isolated proteins and starches, emulsifiers, flavor enhancers, colorants, and preservatives — combined specifically to maximize palatability, shelf stability, and convenience rather than to simply preserve or prepare a recognizable base ingredient. The defining feature isn’t any single nutrient (an ultra-processed food isn’t necessarily higher in sugar, fat, or calories on paper than an unprocessed alternative) — it’s the formulation and manufacturing process itself.
What a controlled NIH trial actually found
A 2019 NIH inpatient trial, published in Cell Metabolism, tested the ultra-processed question directly rather than relying on observational association alone. Twenty adults each spent two weeks eating an ultra-processed diet and two weeks eating an unprocessed diet, in random order, with meals deliberately matched for presented calories, energy density, macronutrients, sugar, sodium, and fiber. Despite that matching, people consumed measurably more calories — and gained weight — during the ultra-processed weeks, while losing weight during the unprocessed weeks. This is a genuinely striking result specifically because the diets were designed on paper to be nutritionally equivalent.
Why matched nutrition didn’t produce matched results
The trial wasn’t designed to isolate one single mechanism, but researchers have proposed several contributing factors: ultra-processed meals in the study were eaten faster, which may have outpaced normal fullness signals; the higher palatability common to ultra-processed formulations may drive more consumption per sitting; and differences in energy density and texture between the two diets likely played a role even at matched macronutrient content. The honest summary is that “eating rate and palatability probably matter, and research is still working out exactly how much each contributes” — not a single, fully settled mechanism.
This site’s food database doesn’t classify by processing level
Every food in this site’s database carries verified per-100g nutrient values — calories, protein, carbs, fat, fiber, and more — checked against USDA FoodData Central, but it doesn’t carry a NOVA-style processing tag, since that’s a different dimension than the nutrient data the underlying source is built around. The ingredient-list heuristic below is a reasonable way to apply the same general idea without a formal classification built into the tool.
A worked comparison: two snacks with similar labels
Consider two snacks with a nearly identical calorie, protein, and sugar breakdown on paper: a piece of fruit with a small handful of nuts, and a packaged granola bar formulated to hit similar numbers. Matched macros don’t guarantee a matched effect — the granola bar is more likely to include emulsifiers, isolated starches, and flavoring designed to maximize how quickly and completely it’s eaten, while the fruit-and-nuts combination requires more chewing, digests differently, and doesn’t share that same formulation goal. This is the practical, food-level version of what the NIH trial above found in a controlled setting: matched numbers on a label don’t necessarily mean a matched real-world effect. For how added sugar is counted on a label and the limits published for it, see added sugar per day: limits in grams.
Beyond weight: other associations in the research
Separate from the controlled weight-gain trial above, a large body of observational research has found associations between higher ultra-processed food intake and various other long-term health outcomes, including cardiovascular disease markers. Observational research like this shows association, not proof of direct cause the way a controlled trial does — people who eat more ultra-processed food may differ from those who don’t in other ways too — but combined with the controlled trial’s more direct evidence on calorie intake and weight, it adds to a reasonably consistent overall picture rather than standing alone. A 2026 example: our look at a new NHANES-based study linking ultra-processed food intake to prostate cancer risk walks through both the headline number and, just as prominently, the real limits of that kind of observational design.
Not all processed food is equal
It’s worth resisting the instinct to treat “processed” as a single bad category. Frozen vegetables, canned beans, and plain yogurt are all technically processed, and none of them resemble what current research is actually concerned about — frozen vegetables in particular hold their nutrition close to fresh and are a genuinely reasonable, practical choice, not a compromise. The concern concentrates specifically at the ultra-processed end of the spectrum, not at “anything that isn’t eaten raw and untouched.”
Common mistakes in thinking about processed food
A few patterns lead to unhelpful conclusions here. Treating “processed” as a single binary category — avoiding anything that comes in a package, including canned beans and frozen vegetables — throws out genuinely useful, nutritious, and often more affordable options over a distinction that doesn’t actually apply to them. Assuming a homemade version of something is automatically better just because it wasn’t industrially produced skips over the fact that a homemade dessert made with the same sugar and fat as a packaged one isn’t meaningfully different nutritionally. And fixating on a single ingredient in an otherwise reasonable food — one emulsifier in an otherwise minimally processed product — tends to miss the bigger picture the overall formulation represents.
Seed oils get caught in this same all-or-nothing thinking, often lumped in with “ultra-processed” as a single ingredient-level red flag. Our dedicated look at the seed oil debate covers why that specific ingredient-focused framing doesn’t hold up well against the actual research on omega-6 and inflammation.
The whole-versus-processed distinction also shows up in research that isn’t about weight at all: a 2026 trial comparing ketogenic, Mediterranean, and plant-forward diets on liver fat found that a plant-forward pattern built around whole foods still improved metabolic markers even though it produced the smallest liver fat change of the three diets tested.
A practical way to think about it, without needing a formal label
A useful, low-effort heuristic: a short ingredient list of recognizable items (or none at all, for a single-ingredient food like a vegetable or a cut of meat) generally sits toward the less-processed end; a long list including unfamiliar additives, isolates, and stabilizers generally sits toward the ultra-processed end. This isn’t a perfect, formal classification — a food additive isn’t automatically harmful just because it’s unfamiliar — but it’s a reasonable everyday proxy that doesn’t require memorizing the full NOVA framework.
Bringing this into a real week of eating
Cooking from base ingredients — the same approach already covered for batch cooking and budget-conscious meal planning — naturally skews toward the less-processed end of the spectrum, simply because home cooking typically starts from ingredients rather than industrial formulations. None of this requires eliminating every packaged food; it’s a reasonable general direction to lean, not a rule to enforce on every single item in a shopping cart.
The goal realistically isn’t zero ultra-processed food — for most people that’s neither necessary nor sustainable — it’s making the base of a normal week’s eating whole and minimally processed foods, with packaged and ultra-processed items filling in around the edges rather than making up the majority of what’s actually eaten.