Coffee and dementia: what the 2026 JAMA study shows and what it doesn’t

Coffee does not reliably protect against dementia. In a large US study, though, two to three cups a day went with slightly fewer dementia diagnoses.Data and limits of the JAMA study, earlier meta-analyses, German medical societies and caffeine guidelines.

The study appeared in the US journal JAMA on 9 February 2026 and was picked up by many media outlets. Researchers in Boston had analysed the drinking habits of around 132,000 people over up to 43 years. High intake of caffeinated coffee went with 18 per cent fewer dementia diagnoses.

That sounds like protection from a cup. Yet the study shows an association, not an effect, and the difference is small in absolute terms. Earlier reviews also came to mixed results.

The study: two US cohorts over up to 43 years

The researchers analysed two well-known long-term studies of health professionals. The Nurses’ Health Study has followed female nurses since 1980, the Health Professionals Follow-up Study male health professionals since 1986. Every two to four years, participants reported how much caffeinated coffee, decaf and tea they drank.

A dementia case meant a doctor’s diagnosis, identified through death certificates and participants’ reports. Alzheimer’s could not be analysed separately. The authors adjusted for many other factors, including age, education, smoking, alcohol, exercise, weight, blood pressure, diabetes, depression and diet quality. No funding from the coffee or tea industry is declared.

The results: 18 per cent relative, about 2 in 100 absolute

Each comparison set the group with the highest intake against the group with the lowest. For caffeinated coffee, the dementia risk was 18 per cent lower. The confidence interval runs from 11 to 24 per cent, so the association is statistically clear. Tea showed a similar pattern, decaf did not.

The 18 per cent is a relative figure. The British statistician Kevin McConway translated it into absolute numbers for the Science Media Centre. In the group drinking the least coffee, about 12 in 100 participants received a dementia diagnosis over the study period. In a comparable group with the highest intake, it would be about 10 in 100.

More did not help more. The risk was lowest at two to three cups of coffee a day, which the study puts at around 300 mg of caffeine. For tea, it was one to two cups. Above that, the curve flattened. In memory tests, run only on 17,139 women aged 70 and over, the differences were very small. Study lead Daniel Wang himself stresses that the effect is small.

The limits: an association is not an effect

An observational study cannot show that coffee is the cause. The authors say so themselves. Independent experts asked by the UK Science Media Centre raise three main objections.

• Reverse direction: Dementia begins years before the diagnosis. In this early phase, people may sleep worse and so drink less coffee or switch to decaf. That fits with heavy decaf drinkers reporting memory problems more often.
• Other factors: Coffee and tea drinkers might live differently overall. Much was adjusted for, but some unmeasured influence may remain.
• Transferability: Almost all participants were US health professionals. ‘A lot of coffee’ is only defined within these groups, and for tea the highest group equals about one cup a day.

The questionnaire recorded neither the type of tea nor the brewing method. So espresso, filter coffee and unfiltered coffee cannot be told apart. Which substances might lie behind the association remains open. The authors name caffeine as a possible factor, alongside polyphenols such as chlorogenic acid in coffee and catechins in tea.

Context: what meta-analyses find on coffee and dementia

The JAMA study is not the first on the subject. Reviews that pool many cohorts have so far reached different results. They too rely almost entirely on observational data, as there are no long-term randomised trials.

Taken together, there is no clear picture. In most data, moderate coffee intake is not linked to more dementia, but protection is not established. In the UK Biobank analysis, very high coffee intake was even linked to a higher risk. That too is an observation and not proof.

Medical societies: coffee is not a prevention measure

The Deutsche Gesellschaft für Neurologie (German Neurological Society) commented on the JAMA study on 27 February 2026. Its expert Peter Berlit considers the effect noteworthy and the results physiologically plausible. At the same time, he stresses that the data are only observational.

On prevention, experts set other priorities. The Lancet Commission lists 14 modifiable risk factors, including high blood pressure, hearing loss, smoking, physical inactivity, diabetes, depression and social isolation. By its estimate, around 45 per cent of dementia cases worldwide could be delayed or prevented. For Germany, an analysis with the German Center for Neurodegenerative Diseases (DZNE) from December 2025 arrives at around 36 per cent.

Coffee is on none of these lists. The Deutsche Alzheimer Gesellschaft (German Alzheimer’s Society) derives ten measures from the 14 factors, and coffee is not among them. It does not yet count diet as a whole among the factors, because the data are not sufficient.

How much coffee a day: the EFSA values

The two to three cups from the study fall within the amounts the European Food Safety Authority considers safe for healthy adults. Germany’s Federal Institute for Risk Assessment (BfR) gives the same values. They do not apply to people with medical conditions or taking certain medicines.

How much caffeine is in your cup depends heavily on the bean and the brewing method. Guide values for espresso, filter and cold brew are in Caffeine in espresso, filter coffee and cold brew: the values per cup. Tea counts too, and for powdered green tea the figures are in How much caffeine does matcha have, and what does L-theanine do?

The JAMA study is large, long and carefully done. It shows that moderate coffee intake is no reason for concern about the brain. Protection against dementia would need studies that can separate cause and effect. If you are worried about your memory, talk to your GP.

Last updated: 24 October 2026.