Does coffee reduce dementia risk? What the 2026 study found, and what it can't tell you

Short answer: Moderate coffee drinking — about two to three cups a day of caffeinated coffee — is consistently associated with a somewhat lower risk of dementia. The largest study to date, published in JAMA in March 2026, followed more than 131,000 people for up to 43 years and found about 18% lower risk in the highest quarter of caffeinated-coffee intake compared with the lowest. Decaf showed no such link, and very heavy drinking (more than six cups a day) is associated with higher risk in other research. Genetic studies designed to test cause and effect have not found that coffee changes cognition. So: if you enjoy coffee, the evidence is reassuring. It is not a reason to start.

This is educational content, not medical advice. It describes associations found in population research, not predictions about any individual.

What the 2026 JAMA study found

The study pooled two long-running American cohorts: the Nurses' Health Study (86,606 women) and the Health Professionals Follow-up Study (45,215 men). Participants reported their diet every two to four years, which matters — it means the analysis tracked habits over decades rather than relying on one questionnaire. Over a median of almost 37 years, 11,033 people developed dementia.

• Caffeinated coffee: highest vs lowest quarter of intake, HR 0.82 (95% CI 0.76–0.89) for dementia. Subjective cognitive decline was also less common (7.8% vs 9.5%).
• Tea: similar associations, most pronounced at about one to two cups a day. (We cover the tea evidence, including green tea, in more depth here.)
• Decaf coffee: no association with dementia risk or cognitive performance.
• The shape: the benefit was not "more is better." The lowest risk sat at about two to three cups a day of caffeinated coffee, with no further benefit at higher intakes. (The study defined a cup as 8 oz, about 240 ml.)
• Thinking tests: in the women who took them, heavier coffee drinkers scored 0.11 points higher on a 41-point telephone cognitive test. On the broader global-cognition score, the difference was not statistically significant.

One point worth getting right, because some coverage got it wrong: "NHS" here stands for the Nurses' Health Study, a US cohort. These were not British National Health Service nurses. Both cohorts are American, and both are made up of health professionals — people who are generally better educated and healthier than average, which limits how far the results generalise.

Other studies point the same way — mostly

That last row is the part of the story that rarely makes the headlines. Across several analyses, the curve bends back up at high intake. Non-drinkers and decaf drinkers also had higher odds than light drinkers in the 2022 UK Biobank analysis — which is one of the reasons researchers think who drinks coffee matters as much as the coffee.

Why "associated with" is doing a lot of work

Three reasons to hold the headline lightly:

1. The genetic studies are null. Mendelian randomisation uses gene variants that make people drink more or less coffee as a kind of natural experiment, less vulnerable to lifestyle confounding. A 2018 analysis of up to 415,530 people across ten European cohorts found no evidence that genetically higher coffee intake affected global cognition or memory — and it had the statistical power to detect very small effects. A 2024 Mendelian-randomisation study likewise found no causal association between coffee and Alzheimer's disease (though it did find one for migraine).

2. Reverse causation. People in the earliest, undiagnosed stages of cognitive decline — or with other health problems — may cut back on coffee. That would make coffee drinkers look healthier without coffee doing anything. Very long follow-up with repeated diet measurements, as in the JAMA study, reduces this problem but cannot rule it out.

3. Coffee drinkers differ. In the JAMA study, heavier caffeinated-coffee drinkers tended to be younger, drink more alcohol, smoke more and eat more calories. Studies adjust for what they can measure, and only as well as they measure it.

The decaf result cuts both ways. If the association were purely about lifestyle, decaf drinkers — who share much of the coffee-drinking lifestyle — might be expected to show it too. They did not, which nudges towards something about caffeine. But decaf drinkers may also be a distinct group (some people switch for health reasons), so it is a clue, not a proof.

If you want the general version of this — why an observational hazard ratio is not a personal promise — see our guide to reading dementia statistics.

So should you drink coffee for your brain?

If you already drink coffee and tolerate it well: the evidence is reassuring. Two to three cups a day sits right in the range associated with the lowest risk.

If you don't drink coffee: there is no good reason to start for brain health. The possible benefit is modest, unproven as cause and effect, and much smaller than the levers you control directly — like blood pressure, hearing and physical activity.

If you drink a lot: in UK Biobank, more than six cups a day is where the association turned unfavourable, and one 2026 meta-analysis saw Alzheimer's risk start to rise above about three cups. For most adults, the FDA cites 400 mg of caffeine a day — about two to three 12-ounce cups — as an amount not generally linked with harmful effects.

Watch what goes in it. A large syrup-and-cream coffee drink is a sugar-sweetened beverage with coffee in it. The swap evidence runs from sugary drinks towards plain coffee and tea, not the other way — more on sugary drinks and dementia.

Protect your sleep. The FDA lists insomnia and sleep disruption among the signs of too much caffeine, and poor sleep has its own links to dementia risk. If coffee affects your sleep, try having it earlier in the day.

Check with your clinician if you have an irregular heartbeat, high blood pressure, anxiety, reflux, are pregnant, or take medicines that interact with caffeine.

Coffee is, at most, a small part of the picture, and one item within your overall eating pattern (see the MIND diet). The free Solenna risk profile takes about ten minutes and shows which modifiable factors are most worth your effort — so the big levers don't get lost behind the small ones.

How we research and review our content →

An independent physician audited the Solenna app. We published all nine findings →

Educational content only. It does not diagnose any condition and is not a substitute for professional medical advice. If you have a heart rhythm problem, high blood pressure, are pregnant, or take regular medication, discuss your caffeine intake with your clinician.