Three servings of fries a week raised diabetes risk by 20 percent. Other potatoes did not
A Harvard analysis of more than 200,000 adults separated fried potatoes from baked, boiled and mashed ones. Only the fries were linked to a higher risk of type 2 diabetes.
Potatoes are often grouped together as one food. A team at the Harvard T.H. Chan School of Public Health looked at how they are cooked.
Seyed Mohammad Mousavi, Walter Willett and colleagues published the study in The BMJ in 2025. They analysed data from 205,107 adults in three long-running US cohorts, followed for up to almost four decades, and combined it with other cohort studies in a separate analysis.
What they found
- Eating three servings a week of French fries was linked to a 20 percent higher risk of type 2 diabetes.
- Baked, boiled and mashed potatoes were not meaningfully linked to diabetes risk.
- Replacing three servings a week of fries with whole grains was linked to a 19 percent lower risk.
- Replacing potatoes with white rice was linked to a higher risk, not a lower one.
What it means
The problem is less the potato and more what happens to it. Frying adds a large amount of fat and calories, fries are often eaten with salty, processed food, and the swap matters as much as the food removed. Whole grains, such as oats, brown rice and wholemeal bread, came out as the better choice.
How to apply it
- Keep fries as an occasional side, not a weekly habit.
- Choose boiled, baked or roasted potatoes with the skin, which keeps more fibre.
- When cutting potatoes, replace them with whole grains or legumes, not white rice or white bread.
- Watch portion sizes and what is on the rest of the plate.
A note on the evidence
The study was observational, so it cannot prove that fries cause diabetes. Diet was self-reported and the cohorts were mostly made up of white American health professionals. The size of the population and the consistency with substitution analyses make the main finding credible.
Sources
- Mousavi SM et al. Total and specific potato intake and risk of type 2 diabetes: results from three US cohort studies and a substitution meta-analysis of prospective cohorts. BMJ, 2025. doi:10.1136/bmj-2024-082121




