A 2026 study of 17,024 US men found that higher ultra-processed food intake was linked to a 24-31% higher rate of self-reported prostate cancer, with one subgroup showing a 34% difference that did not reach statistical significance. The data comes from NHANES, a national US health survey, and the study is observational: it shows an association, not proof that ultra-processed food causes prostate cancer.

What the New Study Found

Researchers at Florida Atlantic University’s Charles E. Schmidt College of Medicine, led by senior author Charles H. Hennekels, analyzed diet and health data from 17,024 US men and published the results in The American Journal of Medicine in 2026. Men were sorted into four groups based on what share of their daily calories came from ultra-processed food, from under roughly 20% to over roughly 44%.

  • Men in the combined higher-intake groups had a 24-31% greater risk of self-reported prostate cancer than men in the lowest-intake group, after adjusting for age, smoking status, race and ethnicity, and poverty status.
  • The single highest-intake group, compared against the lowest on its own, showed a 34% difference that did not reach statistical significance, which the researchers attributed in part to a smaller number of prostate cancer cases within that narrower group.
  • The foods driving classification into the ultra-processed category included soft drinks, packaged snacks, sweetened breakfast cereals, and processed meats.

That range, roughly 24% to 34% depending on exactly which groups are compared, is the headline number behind the recent wave of coverage. It is a real, adjusted statistical finding from a large national sample, and it is also just one part of the picture. The next section covers who was actually studied and how the researchers arrived at that number.

Advertisement

Who Was Studied, and How

The study drew on NHANES, the National Health and Nutrition Examination Survey, a long-running federal health survey that recruits a nationally representative sample of US adults on an ongoing basis. This analysis pooled data collected between 2003 and 2023, covering 17,024 men ages 18 and older.

To measure ultra-processed food intake, researchers used two 24-hour dietary recalls per participant, where a trained interviewer asks someone to report everything they ate in the previous day. Each food reported was sorted into one of the four NOVA classification groups, and researchers calculated what share of each participant’s daily calories came from NOVA group 4, the ultra-processed category. Prostate cancer status came from participants’ own reports during the NHANES interview, not from a medical chart review or cancer registry match.

The resulting four groups ranged from under roughly 20% of daily calories coming from ultra-processed food in the lowest group to more than roughly 44% in the highest. Worth sitting with for a moment: even the “low” comparison group in this study was already drawing a fifth of its calories from ultra-processed sources. This wasn’t a comparison between people who ate no ultra-processed food and people who ate a lot of it; it was a comparison between a lot and even more, which is a more realistic reflection of how most people in the US actually eat than a hypothetical zero-ultra-processed baseline would be.

Study elementDetail
Sample size17,024 US men, ages 18+
Data sourceNHANES, 2003-2023
Dietary measureTwo 24-hour dietary recalls per participant
Classification systemNOVA (four processing-level groups)
Cancer outcomeSelf-reported prostate cancer diagnosis
Study designCross-sectional, observational
Adjusted risk range24-31% higher (statistically significant)
Highest-intake group vs. lowest34% higher (not statistically significant)
PublicationThe American Journal of Medicine, 2026

This design has real strengths. A sample of over 17,000 men, drawn through NHANES’s established national sampling methods, is large and broadly representative of the US adult male population, which is not something a small clinical study can offer. It also has real limits. The section below covers them plainly, as prominently as the headline number itself, instead of tucking them in at the end.

What Counts as “Ultra-Processed,” Quickly

NOVA sorts food into four groups by how much industrial processing and formulation they involve, not by traditional nutrient categories like calories or fat content. Group 4, ultra-processed food, covers industrial formulations built from ingredients rarely used in home cooking: isolated proteins and starches, emulsifiers, flavor enhancers, and preservatives, combined specifically to maximize shelf life, palatability, and convenience. Soft drinks, packaged snacks, sweetened cereals, and many processed meats fall into this group.

This article doesn’t re-explain the full NOVA framework here; our dedicated look at whole versus processed foods covers the four NOVA groups in depth, including a controlled NIH trial that tested ultra-processed intake directly against calorie intake and weight. That piece is the background reading for the classification system itself; this one is about what a specific new study found when it applied that same system to cancer risk.

What This Study Can’t Tell You

This is the part of the coverage that matters as much as the headline number, and it deserves the same weight here.

The study is cross-sectional and observational. Researchers measured diet and prostate cancer status at roughly the same point in time, in a survey that was never designed to test this specific question. That design can identify a statistical association between two things; it cannot establish that one caused the other, and it cannot rule out that some third factor, unmeasured here, explains part or all of the link.

Prostate cancer status was self-reported. Participants told NHANES interviewers whether they had been diagnosed with prostate cancer, and there was no independent verification against a medical record or a cancer registry. Self-reported diagnosis is a reasonable tool for a survey this size, but it is a real source of potential error compared with a confirmed clinical diagnosis.

Diet was captured from just two 24-hour recalls per person. Two days of reported eating is a standard, practical way to estimate diet across a large sample, but it’s a narrow window used to represent someone’s long-term eating pattern, and it depends on people accurately remembering and reporting what they ate.

Advertisement

The researchers’ own highest-risk comparison did not reach statistical significance. The 34% figure for the single highest-intake group against the lowest is the number that generated the most striking headlines, and it’s also the one the study’s own authors flagged as not statistically significant, most likely because fewer prostate cancer cases fell into that narrower group. The 24-31% range, based on broader group comparisons, is the more statistically solid part of the finding.

There’s also a framing point worth understanding on its own, separate from the design limits above. A 24-34% figure describes a relative difference: how much higher the rate was in the higher-intake group compared with the lower-intake group, not the actual share of men in either group who reported a prostate cancer diagnosis. Relative differences like this are the standard way research reports this kind of comparison, and they aren’t wrong to use, but a relative increase can read as more dramatic than the same finding would look expressed as a change in absolute numbers. Keep that in mind when a headline states the percentage on its own.

None of this means the study is worthless. It means the honest read is “a real association worth taking seriously and worth further study,” not “ultra-processed food has been shown to cause prostate cancer.” The researchers themselves called for larger observational studies and randomized trials to test the hypothesis further, which is a more measured conclusion than most headlines about this study conveyed. This gap between a striking headline and a more careful reading of the underlying evidence isn’t unique to this study; our look at whether seed oils are actually harmful walks through a different food-science claim where the same kind of gap shows up.

How This Fits With the Wider Ultra-Processed-Food Evidence Base

This study doesn’t stand alone. A separate body of research, including a controlled NIH inpatient trial, has already linked higher ultra-processed food intake to excess calorie intake and weight gain under tightly controlled conditions. That trial design carries more weight than an observational survey for one specific reason: it randomizes what people eat instead of only measuring what they report eating, which removes a lot of the guesswork about which direction the effect actually runs. Other, separate observational research has associated higher ultra-processed intake with cardiovascular disease markers and other long-term health outcomes.

Prostate cancer risk specifically is a newer, thinner line of evidence within that broader picture. A separate systematic review and meta-analysis of earlier research on ultra-processed food and prostate cancer risk did not find a significant association overall, though it noted a possible link to prostate cancer mortality specifically. That’s a useful reality check: this new NHANES-based study adds a large, adjusted, nationally representative data point to a mixed evidence base, not a settled answer that closes the question.

Put together, the pattern across all of this research points toward ultra-processed food intake as a factor worth reducing for reasons that already have firmer evidence behind them, weight and cardiovascular markers among them, while prostate cancer risk specifically stays an active, developing area rather than a confirmed one.

Practical Ways to Lower Ultra-Processed Intake Without an Overhaul

None of this calls for eliminating every packaged food or treating a trip to the grocery store as a minefield. A more useful frame is shifting the base of a normal week’s eating toward whole-food versions of the same nutrients, protein in particular, since protein-focused ultra-processed products (protein bars, flavored shakes, breaded and reformed meats) are common and easy to swap.

Small, repeatable substitutions add up more reliably than an ambitious short-term overhaul that’s hard to sustain past a few weeks. The goal here isn’t a perfect week; it’s shifting the everyday default a little, in a direction the wider evidence already supports for other reasons.

  • Swap a packaged snack for a whole-food option with a similar role: fruit and a handful of nuts instead of a granola bar, plain yogurt instead of a flavored one.
  • Choose whole cuts of meat, fish, eggs, legumes, or tofu over reformed or breaded protein products when planning meals. Our full breakdown of protein sources, ranked by density, cost, and versatility, is a practical starting point.
  • Check ingredient lists as a quick heuristic: a short list of recognizable items generally sits toward the less-processed end; a long list of unfamiliar additives and isolates generally sits toward the ultra-processed end.
  • Build fat sources around whole foods, olive oil, nuts, and fatty fish, in place of industrial spreads and fried packaged snacks. Our explainer on healthy fats covers which sources to lean on.
  • Track where protein and calories are actually coming from over a week, instead of judging any single meal in isolation. The protein calculator below is a starting point for figuring out a personal target and checking how a day’s actual eating measures up against it.

None of these are framed as foods to fear or avoid entirely. They’re substitutions that move the overall shape of a week’s eating in a direction that already has firmer evidence behind it on weight and cardiovascular markers. Prostate cancer risk is layered on top as a newer, still-developing reason; it isn’t the sole justification for any of them.

This article covers population-level research, not an individual risk assessment. Anyone with specific concerns about prostate cancer risk, family history, or when to start screening should raise that with a doctor, who can weigh a person’s actual history in a way a single study never can.