Older adults who ate the most ultra-processed foods had a 58 percent higher risk of developing dementia over nearly nine years, compared with peers who ate the least. Those who leaned hardest on minimally processed whole foods had a 41 percent lower risk.
The finding comes from an analysis of 5,370 US adults over 50 in the Health and Retirement Study, published in the American Journal of Public Health in June 2026. Participants averaged 64 years old at baseline and took standardized memory and attention tests every two years.
The pattern is not limited to the oldest group. In a separate 2025 Framingham Heart Study analysis, each extra daily serving of ultra-processed food in middle age was linked to a 13 percent higher risk of Alzheimer's disease. Among participants under 68 at baseline, those eating 10 or more servings a day had almost triple the Alzheimer's risk of those eating fewer.
What counts as ultra-processed
The label comes from the Nova classification, developed by researchers at the University of São Paulo. It sorts foods by how much industrial processing they undergo, from whole foods like fruit and grains at one end to ultra-processed formulations at the other.
Also read: Your standing desk may not protect your heart, and it could raise a different risk
Ultra-processed products are industrial mixes built from extracted substances such as sugars, oils, starches and protein isolates, then recombined with additives like flavors, colors and emulsifiers. Typical examples include soft drinks, packaged snacks, chicken nuggets, instant soups, sweetened cereals, fruit drinks and flavored yogurts.
In the United States and United Kingdom, these products now supply more than half of daily calories, according to the Australian research team behind the most recent cognitive study.
What the wider evidence shows
A 2024 umbrella review in The BMJ pooled 45 meta-analyses covering nearly 10 million people. It tied higher ultra-processed food intake to 32 negative health outcomes.
The strongest links were with cardiovascular disease death, common mental disorders, obesity and type 2 diabetes. All four are themselves recognized risk factors for cognitive decline later in life.
Also read: 6 evidence-backed habits that lower cortisol, and may protect your long-term health
A cross-sectional study of 2,192 Australian adults aged 40 to 70, published in April 2026, added another piece. Each 10 percent rise in the share of calories from ultra-processed foods was tied to lower attention scores and a higher estimated dementia risk, even after adjusting for overall Mediterranean-style diet quality.
These are observational studies. They cannot prove that ultra-processed foods themselves cause cognitive decline, and the Harvard team was explicit that its results do not establish causation. But the direction across cohorts is consistent.
Cindy Leung, associate professor of public health nutrition at the Harvard T.H. Chan School of Public Health, noted that the analysis also found lower risks of cognitive impairment and dementia among people who ate the most minimally processed foods such as whole grains, fruits and vegetables. The overall pattern, she said, suggests "there may not be a safe level."
Small swaps that fit the evidence
The research points to a straightforward direction. Trade a sweetened breakfast cereal for oatmeal with fruit. Replace processed meats, which drove the strongest risk signal in the Harvard analysis, with beans, eggs or plain chicken.
Also read: New study raises concerns: Pregnant women exposed to an average of 45 chemicals
Choose water or unsweetened coffee over soda, and cook more meals from basic ingredients instead of reheating frozen dinners. None of these moves require a special diet, and each one shifts the daily balance toward the foods linked to lower cognitive risk.
Anyone with memory concerns, or worried about a family history of dementia, should raise it with a primary care doctor rather than making dietary changes alone.
Also read: Walking more throughout the day may undo much of the harm from hours of sitting