Packages of plant-based meats, which are classified as ultra-processed, at a grocery store in 2025. | Bloomberg via Getty ImagesIn little more than a decade, the term “ultra-processed foods” (UPFs) has risen from an obscure academic coinage to one of the most potent ideas in the American food imagination. It has saturated media coverage of diet and disease, spawned a profusion of guides teaching shoppers how to spot UPFs at the supermarket, and animated Robert F. Kennedy Jr.’s crusade to remake American food policy. It might also be kind of fake. The trouble starts with the definition. UPF generally refers to packaged foods with questionable-sounding ingredients not typically used in household kitchens (high-fructose corn syrup, maltodextrin, and the like). But not even nutrition scientists can really tell you where normal processing ends and “ultra-processing” begins, and the difference often comes down to vibes. (I once covered a study that, inexplicably, classified tofu as ultra-processed.) Further, much of the evidence linking ultra-processed foods to poor health outcomes such as heart disease and cancer is notoriously low-quality because it’s based on big, noisy observational studies that can’t disentangle correlation from causation. That weakness plagues a lot of nutrition research, but it’s especially notable for UPF studies, because many of them are drawn from diet surveys that don’t capture enough detail to tell whether the “white bread” or “yogurt” someone reported eating was ultra-processed in the first place.To correct for those problems, a handful of nutrition researchers have run randomized controlled trials (RCTs) — a gold standard for establishing cause and effect — to isolate the health effects of “ultra-processed” foods. Several of the best-known trials have found that UPFs indeed do contribute to adverse health outcomes such as overeating and weight gain — seemingly offering the stronger evidence that observational studies could not. But it turns out those trials might not show what they claim to, either. In a recent article published in Science, a group of researchers analyzed data from five landmark randomized controlled trials of ultra-processed foods and found that most of them compared ultra-processed and non-ultra-processed diets that differed in important ways beyond processing. The ultra-processed diets tended to be more calorie-dense and lower in fiber, for example, which can lead to overeating regardless of whether a food is “ultra-processed.” Accordingly, some of the effects that they attributed to processing may have had other causes.These findings matter well beyond an academic fight over food categories. They offer the latest evidence against taking the UPF label too seriously when deciding what to buy — or fear — at the grocery store. Randomized controlled trials sought to add clarity to the UPF debateUPF researchers sort foods using NOVA, a classification system with four tiers. From least to most processed:Group one includes unprocessed or minimally processed foods, like whole fruits, vegetables, whole grains, beans, nuts, milk, and cuts of meat.Group two encompasses “processed culinary ingredients,” like cooking oils, butter, sugar, and salt. Group three is made up of processed foods produced by combining group one and group two ingredients into things like homemade breads, desserts, sautés, cured meats, and other dishes. Group four, or ultra-processed foods, is defined as “formulations of ingredients, mostly of exclusive industrial use, that result from a series of industrial processes,” including dyes, preservatives, emulsifiers, and other ingredients rarely or never found in home kitchens.The categories sound tidy. They are not. Many nutrition experts have criticized them as overly broad and hard to apply consistently. Wheat gluten, for example, is an ingredient often added to packaged breads; it’s not used in many household kitchens, and is usually classified as ultra-processed. Yet it’s been used in many Asian cuisines for centuries and can easily be employed in your home kitchen; I use it in my own. There’s no evidence that the ingredient’s rarity in household pantries makes it unhealthy or deserving of a special “ultra-processed” label. Nor is there any reason to assume that a store-bought whole grain bread with added gluten (category 4 under NOVA) is worse for you than, say, a homemade white bread (category 3).The best ultra-processed food to eat on the 4th of JulyLast Independence Day, I (okay, my husband) made these life-changing plant-based ribs from Miyoko Schinner’s The Homemade Vegan Pantry. The magic ingredient is wheat gluten, which the NOVA classification treats as a marker of “ultra-processing,” but we’ll be repeating them every Fourth until I die. The full recipe is available on Schinner’s YouTube channel here.To be fair to the UPF framework, its whole point is to try to capture something distinct from traditional nutritional metrics, to help make scientific sense of why modern, industrially produced food feels wrong. Some researchers argue that reducing foods to their individual nutrients misses something about industrial food as a whole. But many of the other factors that can make foods unhealthy or prone to being overeaten — calorie density, soft texture, hyper-palatability — are already well-known to nutrition science. The question is whether “ultra-processing” explains anything beyond those familiar mechanisms.Kevin Hall, one of the most prominent nutrition scientists in the US, known for his research on UPFs, told me in an email that he was initially skeptical of the NOVA system but has come to appreciate some of its attributes as “a complementary lens for viewing food that can be a helpful integrative addition to the reductionist nutritional profiling lens.” Researchers like Hall have employed randomized controlled trials to rigorously test whether “ultra-processing” has real, and really troubling, effects. These trials work by recruiting a few dozen participants and assigning them pre-specified diets that are similar in their nutritional profiles but different in processing level. The researchers then examine whether any differences in participants’ health outcomes, like weight gain and cholesterol levels, can be pinned on the ultra-processed diets. Hall’s seminal RCT, one of the five trials examined in the new Science analysis, controlled for things like the number of calories made available to participants on unprocessed and ultra-processed diets; the amount of carbs, fat, and protein in those diets; and other factors. It found that the ultra-processed diets led to increased calorie intake and weight gain. The Science analysis, though, finds that the ultra-processed and non-ultra-processed diets in Hall’s study and other RCTs still differed in ways that could confound the trials’ results. The UPFs used in the trials tended to be softer than the non-ultra-processed foods, which can accelerate eating rate and lead to excess calorie consumption. They also tended to be more energy-dense (meaning that they had more calories per gram of food), less fiber-dense, and higher in saturated fat and sodium. These aren’t inherent properties of “ultra-processed” foods. “While UPFs are often characterized by properties such as soft texture, higher energy density, and lower fiber content, these features are neither unique to UPFs nor consistently present across them,” Faidon Magkos, a professor in obesity and metabolism at the University of Copenhagen and an author of the Science analysis, told me in an email. “There are many non-UPF foods with similar properties, and conversely, many UPFs that are harder in texture, lower in energy density, or higher in fiber.” (To name one example, I eat these ultra-processed bars, which are hard in texture and a good source of fiber, multiple times a week.) A core issue with most of the trials is that they compared ultra-processed diets (NOVA group 4) with diets made mostly or entirely of unprocessed foods (group 1). That “extreme contrast,” as Magkos and his co-authors call it, is not the cleanest comparison to make if the goal is to isolate what ultra-processing itself does. Nutrition scientists already know that unprocessed diets made up of whole foods are generally the healthiest for us and that many age-old processed foods, like white flour, white sugar, and cured meat, can be bad for us. But the question is whether there’s something uniquely bad about ultra-processing when compared with other types of processing. Consider a pair of meals from one of the trials: One ultra-processed breakfast consisted of Honey Nut Cheerios, milk, a powdered fiber supplement, a packaged blueberry muffin, and margarine. An unprocessed breakfast, meanwhile, included whole oats, blueberries, almonds, milk, and salt. These are so different that it would be hard to attribute the health differences to the presence of, say, tripotassium phosphate in Honey Nut Cheerios, rather than to the added sugars and refined flour in the ultra-processed meal — long-established nutrition problems that can be explained without the UPF framework. When I asked Hall about the criticisms raised in the Science analysis, he wrote that the analysis “presents plausible hypotheses to explain the observations from past UPF studies” but argued that its “implication seems to be that we already understand enough about the past UPF data and can simply skip the step of critically testing our ideas with careful experiments.” Magkos and his co-authors, for their part, do call for more research refining the past trials and better isolating the specific attributes of UPFs. This is not just academicOne of the most insightful critiques I’ve heard of the focus on ultra-processed foods came from Kevin Klatt, a nutrition scientist at the University of Toronto. The definition of ultra-processed foods in the NOVA framework, he told me in a conversation last year, “has absolutely nothing to do with processing.” The comment surprised me, but Klatt makes a good point: Consider something like edamame pasta, he said. It undergoes intense industrial processing — it’s pulverized and extruded, which changes the plant’s cellular structure and makes its proteins and starches more digestible. Yet it would be categorized by NOVA as minimally processed, because its only listed ingredient is “edamame.” Seen in that light, the problem with UPF as a concept may not be just that it’s hard to define, but that it lacks a reasoned organizing principle beyond casting aspersions on an arbitrarily selected set of ingredients that have little in common with one another. Even so, you might say, if many if not most UPFs — things like Oreos, Pringles, Twinkies — are junk, why does it matter? It matters because nutrition guidance ought to be based on real information about the world, and when it isn’t it leads to bad politics and policy that don’t solve problems and may well make them worse. We have a live case study: a health secretary urging Americans to eat lots of red meat and full-fat dairy, while stigmatizing plant-based diets and “ultra-processed” meat and dairy alternatives. The public’s turn against UPFs, as Vox has written before, has likely contributed to the declining popularity of plant-based meat and milk, even though those foods are often healthier than their animal-based counterparts and are undeniably better for the climate and animals. Federal and some state officials are now seeking to build food regulations around the “ultra-processed” concept for programs like school meals, which could push food companies to reformulate products without making them any healthier. A growing number of programs now exist to certify foods “non-ultra-processed,” and, as Washington Post food columnist Tamar Haspel has pointed out, that has had the effect of lending the halo of health to conventionally, boringly unhealthy packaged foods like cookies. It’s hard to see how this helps consumers make better choices. Meanwhile, they can be misled into making worse choices if the UPF frame nudges them to choose, say, butter instead of a vegetable oil-based spread lower in saturated fat, because they’ve been convinced that the former is healthier (the American Heart Association and other authorities recommend the latter).It’s not hard to see why the idea of ultra-processed foods has resonated so deeply. It taps into a justified feeling that food corporations profit at the expense of our health, and it has quite explicitly tried to turn that instinct into a scientific agenda. But a framework that cannot clearly diagnose what makes our food system harmful offers no real leverage against the industry that profits from it.