Green Tea and Dementia: What Twenty Years of Japanese Research Has Turned Up

In January 2025, a team of Japanese researchers published something... I have not been able to stop thinking about. They had taken MRI scans of 8,766 people aged 65 and over, none of whom had dementia, and gone looking for a pattern. Did the brains of tea drinkers look any different?

They did. The more green tea someone drank, the fewer small areas of damage showed up in their brain's white matter.¹

Then the researchers ran the same analysis on coffee, and found nothing at all.

I sell Japanese tea for a living, so read me with that in mind. But I went into this expecting a thin pile of studies held together by wishful thinking, and that is not what I found.

 Researchers in Sendai, Kanazawa, Niigata and Nagano have spent up to twenty years quietly following the same people, noting down what they drink and waiting to see who develops dementia. One team spent a year giving pensioners capsules of matcha and capsules of coloured cornstarch without telling them which was which. Another worked out what happens to the tea after it leaves your stomach, and the answer involves your gut bacteria doing something I did not see coming.

Here is what they found.

Think of the brain as a building. The grey matter is the rooms, where the work happens. The white matter is the cabling running between them, the wiring that lets one part talk to another.

As we age, that cabling frays. Small blood vessels supplying the white matter narrow and leak, and little patches of damage accumulate. On an MRI they show up as bright spots, and radiologists call them white matter lesions. Almost everyone over 60 has some. They usually cause no symptoms whatsoever.

What they do is predict. People with more of them are more likely to develop dementia later, both the vascular kind and Alzheimer's. So they work as a kind of early warning system, measurable in people who are still completely well, years before anything goes wrong.

Which is why the 2025 study is interesting. It was not measuring dementia. It was measuring the wiring, in people who were fine, and finding that the heavy tea drinkers had less fraying.

The effect was modest. Someone drinking around three cups a day had roughly 3 per cent less lesion volume than a one-cup drinker, and someone drinking seven or eight cups had about 6 per cent less. It also showed up in white matter only. Hippocampal volume and total brain volume were unchanged, so this is not a story about tea keeping your brain bigger.

And the study took its measurements at a single moment. It is a photograph, not a film. It shows that tea drinkers and non-tea-drinkers looked different in 2018, not that the tea did it. The researchers say as much, and say the next step is following people over time.

But the finding fits into something larger, which is where it gets compelling.

Japanese researchers have been circling this question since the mid-2000s, and the studies have got steadily better as they have gone.

The first one, in 2006, looked at 1,003 people aged over 70 in Sendai and found that green tea drinkers scored higher on cognitive tests.² Again, a photograph. It could not rule out the obvious alternative explanation, which is that people whose minds are already slipping stop bothering to make tea.

The Nakajima Project fixed that by following people forward. Starting in 2007, researchers tracked 490 people aged over 60 for an average of five years, and the daily green tea drinkers turned out to be about a third as likely to develop cognitive decline as the non-drinkers.³ Coffee did nothing. Black tea did nothing. Only green tea.

One number from that study has since taken on a life of its own. For dementia specifically, the researchers reported a figure that looks like a 74 per cent reduction in risk, and you will find it quoted all over the internet as exactly that. It should not be. Only a handful of those 490 people developed dementia over the five years, and when you are working with a handful of cases, a result that striking could just as easily be luck. The researchers said so themselves. Their own margin of error was wide enough to include the possibility that green tea did nothing at all. The broader finding, covering the much larger number of people who showed any cognitive decline, is the one that holds up.

Then the sample sizes got serious. The Ōsaki study followed 13,645 older people and found that those drinking five or more cups a day had about a quarter lower risk of developing dementia.⁴ The Saku study found the benefit at two to three cups a day.⁵ A British study, working with 377,592 people in the UK Biobank, found tea drinkers 16 per cent less likely to develop dementia, with the effect strongest for the vascular kind at around 25 per cent.⁶ That last one was almost entirely black tea, which makes it a different question, but the vascular emphasis lines up neatly with those white matter scans.

Now the good part.

In Niigata, researchers followed 13,757 people in the Murakami region and published their results twice.

The first analysis came out in 2021, after eight years. Green tea showed nothing.⁷ No overall association with dementia at all. Coffee, on the other hand, looked terrific: the heaviest coffee drinkers had roughly half the dementia risk of the lightest.

The second analysis came out in 2025, after twelve years, on the same people.⁸ This time green tea did show up, with the heaviest drinkers at about a quarter lower risk, and a measurable step down in risk for every extra 150ml (5 fl oz) cup per day.

Same town. Same participants. Same research group. Four more years of waiting, and a null result became a positive one.

That pair of results tells you something neither study could tell you alone. Whatever green tea is doing to the ageing brain, it is doing it at roughly the speed of erosion. You cannot watch a cliff wear back. You can only measure where it was and come back much later to measure again, and if you come back too soon you will conclude that nothing is happening.

Eight years was too soon. Twelve years was enough.

That one fact explains most of the untidiness in this field. It is why one study finds a benefit at two cups a day and another at five. It is why a result appears in one age group and not another. Everyone is trying to photograph something that moves very slowly, and small differences in when you point the camera change what you get.

It also means nobody has coffee worked out either. The Niigata team found coffee looked excellent for dementia risk. The brain scan study found coffee did nothing at all. Both are good studies. They disagree, and until someone resolves it, anyone telling you confidently which of the two drinks is better for your brain is guessing.

All of the above is observational. People who drink a lot of green tea differ from people who don't in a hundred ways, and no statistical adjustment fully fixes that. The way to settle it is to assign people at random.

In 2024, a group did.⁹ They took 99 people aged 60 to 85, all with mild cognitive impairment or memory complaints, and gave half of them 2g of matcha a day in capsules and half of them coloured cornstarch. Neither the participants nor the researchers knew who had which. This ran for a full year. For scale, 2g of matcha is about what you would whisk into a single bowl of usucha, carrying roughly 170mg of catechins.

The main cognitive tests showed no difference.

I think that result deserves more attention than it gets, and not because it is discouraging. A year is nothing against a signal that took the Murakami researchers twelve years to see. Ninety-nine people is small. And everybody in the trial already had cognitive symptoms, which is rather late to start.

What did move was odd and rather charming. The matcha group got measurably better at reading emotion in people's faces. Sleep quality drifted in the right direction without quite getting there. Neither of those was what the trial set out to test, so they are leads rather than conclusions, but they are strange enough to be worth chasing.

This is my favourite part of the whole subject, and I should say upfront where it comes from. Everything in this section was worked out in mice, and in human cells growing in dishes, mostly by Professor Unno Keiko's group at the University of Shizuoka's Tea Science Research Center. Nobody has watched any of it happen inside a living human brain, because there is no way to do that without opening one. So treat this as the best available theory of why the population studies keep finding what they find, rather than as established fact.

It is a very good theory.

Catechins are the compounds that make green tea taste astringent, and they make up somewhere between 10 and 18 per cent of a dry tea leaf. The one that gets all the attention is epigallocatechin gallate, mercifully shortened to EGCG, which accounts for over half of them.

For any of this to matter, EGCG has to get into your brain. And the brain is guarded by the blood-brain barrier, a lining around its blood vessels that works like an extremely selective bouncer. Almost nothing circulating in your blood gets past it.

Caffeine strolls straight through. In laboratory models, roughly 13 per cent of it crosses in half an hour, which is why coffee hits you so fast. EGCG manages about 4 per cent.¹⁰ It gets in, but it has to squeeze.

Here is the part I did not expect. Most of the EGCG in your tea never reaches your bloodstream at all.

A small fraction is absorbed in the small intestine and reaches the brain two or three hours after you drink. In mice, that arrival switches on a cluster of genes involved in building new connections between neurons, and in a dish, it makes nerve cells sprout new branches. That is the fast delivery.

The rest of the EGCG carries on down into the large intestine, where your gut bacteria take it apart.

They do not destroy it. They convert it. Eight hours or more after your cup of tea, one of the resulting compounds is absorbed from the large intestine, crosses the blood-brain barrier slightly more easily than EGCG itself did, and appears to do much the same job.

So a cup of sencha delivers in two shipments. Express in the afternoon, and slow freight arriving that evening, assembled by subcontractors living in your colon. Once you count the second shipment, the amount of usable catechin your body gets from tea is nearer 40 per cent than the 2 to 8 per cent researchers assumed for years.

Two consequences follow, and both are more interesting than the usual advice.

The first is that drinking enormous quantities is probably not the answer. The human studies hint at this without quite saying it. The Saku researchers found their benefit at two to three cups a day and called it moderate consumption. In the brain scan study, going from three cups a day to seven or eight roughly doubled the tea and did not come close to doubling the effect. And in the mouse work, a low dose of catechins extended lifespan while higher doses did not do more, with the largest dose tested producing the smallest result.¹¹

Mice are not people, and I would not build advice on that last finding alone. But nothing anywhere in this literature supports the idea that you should drink as much as you physically can, and several things gently suggest the opposite.

The second is that this varies enormously between people, and the reason is your gut. Since so much of the effect depends on bacterial conversion, two people drinking identical cups of tea can end up with quite different amounts of the active compound. Anyone who tells you the correct number of cups is making it up.

I find this genuinely delightful. The health benefit of green tea, if it is real, is partly not a property of the tea at all. It is a collaboration.

Two things worth knowing, and I will keep them short because we cover the chemistry properly in our article on catechins.

The first is that hot water gets you catechins and cool water does not. Above 80°C (176°F) catechins dissolve readily, while the sweet, savoury amino acids come out from around 50°C (122°F).¹² So the brewing that extracts the most catechin is also the brewing that makes the most astringent cup. Gyokuro at 50°C is one of the great pleasures in Japanese tea and a poor way to get catechins into you. Cold brew is wonderful in Singapore's heat and extracts relatively little. You cannot have both at once, and I would rather say so than pretend.

The second sorts the teas themselves. Catechins are made by sunlight, so the more sun a leaf sees, the more it carries. Summer-picked leaf beats spring's first flush. Shaded teas carry less, which is why gyokuro, kabusecha and matcha all sit lower than an ordinary sencha despite costing a great deal more. Roasting knocks it down further, so hōjicha is at the bottom.

There is a pleasing irony there. If catechins are what you want, the teas to reach for are the plain everyday ones, not the ceremonial ones. A decent sencha or tamaryokucha beats a luxury gyokuro on this particular measure every single time.

This one surprised me enough that I went looking for the numbers.

Bottled green tea contains considerably less catechin than tea you brew yourself. When Kyoto Prefecture's consumer testing centre compared them, sencha brewed in a kyūsu averaged around two and a half times the catechin content of bottled green tea.¹³ Consumer testing in Hokkaido found most bottled green teas landing somewhere around 30 to 40mg of catechins per 100ml.¹⁴

Some of that is simple dilution, because bottled tea is brewed weak to keep it drinkable cold and to stop it turning cloudy or bitter on the shelf. Some of it is processing, since manufacturers use lower extraction temperatures and add antioxidants like vitamin C to keep the colour from browning, and the catechin suffers for it.

There is one funny exception. The bottled teas sold in Japan as fat-reduction health drinks, the ones with the official tokuho designation, are deliberately fortified and typically carry more than 80mg per 100ml, which puts them well above ordinary bottled tea.¹⁴ If you buy a bottle of green tea in a Japanese convenience store, the one making health claims on the label genuinely does have more catechin in it than the one that isn't.

But if catechins are the point, the answer is a teapot.

We sell a tamaryokucha called EGCG Green Tea that we had laboratory tested, so its EGCG and total catechin content are measured numbers rather than a guess. That is unusual in this industry, and it is the tea I would point you towards if this article has convinced you of anything. It is also, plainly, the thing I make money from.

So here is my position, for whatever a tea seller's position is worth to you.

The evidence is not conclusive, and nobody should tell you it is. But six research groups working independently, across hundreds of thousands of people and two decades, keep finding the same modest signal pointing the same way. There is a coherent biological story underneath it involving a delivery system nobody would have guessed at. And the newest work is starting to suggest where the effect might actually live, in the small blood vessels of the brain rather than anywhere exotic.

That is a genuinely promising body of research, and I am more excited about it now than I was before I started reading.

I would add one thing, not as a disclaimer but because it is true and useful. The factors researchers credit for the recent fall in dementia rates in both Japan and Singapore are not smoking, keeping blood pressure and diabetes in check, staying active, and staying connected to other people. Those are established in a way that tea is not. Green tea belongs alongside them rather than in place of them.

But it is a cheap, safe, and extremely pleasant habit with twenty years of encouraging research behind it and a plausible mechanism underneath. I drank three cups while writing this. I would have drunk them anyway. It is nice to think they might be doing something.

If you are worried about your own memory, or someone else's, please talk to a doctor rather than a tea merchant. Early assessment genuinely changes what can be done. Go with a cup of tea in hand, by all means. Just go.

¹ Shibata S, Noguchi-Shinohara M, Shima A, et al. "Green tea consumption and cerebral white matter lesions in community-dwelling older adults without dementia." npj Science of Food. 2025;9:2. https://doi.org/10.1038/s41538-024-00364-w

² Kuriyama S, Hozawa A, Ohmori K, et al. "Green tea consumption and cognitive function: a cross-sectional study from the Tsurugaya Project." American Journal of Clinical Nutrition. 2006;83(2):355-361. https://pubmed.ncbi.nlm.nih.gov/16469995/

³ Noguchi-Shinohara M, Yuki S, Dohmoto C, et al. "Consumption of green tea, but not black tea or coffee, is associated with reduced risk of cognitive decline." PLoS One. 2014;9(5):e96013. https://doi.org/10.1371/journal.pone.0096013

⁴ Tomata Y, Sugiyama K, Kaiho Y, et al. "Green Tea Consumption and the Risk of Incident Dementia in Elderly Japanese: The Ohsaki Cohort 2006 Study." American Journal of Geriatric Psychiatry. 2016;24(10):881-889. https://doi.org/10.1016/j.jagp.2016.07.009

⁵ Koreki A, Nozaki S, Shikimoto R, et al. "A longitudinal cohort study demonstrating the beneficial effect of moderate consumption of green tea and coffee on the prevention of dementia: The JPHC Saku Mental Health Study." Journal of Alzheimer's Disease. 2025;103(2):519-527. https://doi.org/10.1177/13872877241303709

⁶ Hu HY, Wu BS, Ou YN, et al. "Tea consumption and risk of incident dementia: A prospective cohort study of 377 592 UK Biobank participants." Translational Psychiatry. 2022;12:171. https://doi.org/10.1038/s41398-022-01923-z

⁷ Matsushita N, Nakanishi Y, Watanabe Y, et al. "Association of coffee, green tea, and caffeine with the risk of dementia in older Japanese people." Journal of the American Geriatrics Society. 2021;69(12):3529-3544. https://doi.org/10.1111/jgs.17407

⁸ Kaise R, Kitamura K, Watanabe Y, et al. "Green tea consumption and dementia risk in community-dwelling Japanese people aged 40-74 years: A 12-year cohort study." Journal of Nutrition, Health and Aging. 2025. https://doi.org/10.1016/j.jnha.2025.100615

⁹ Uchida K, Meno K, Korenaga T, et al. "Effect of matcha green tea on cognitive functions and sleep quality in older adults with cognitive decline: A randomized controlled study over 12 months." PLoS One. 2024;19(8):e0309287. https://doi.org/10.1371/journal.pone.0309287

¹⁰ 海野けい子.「緑茶カテキンが脳の老化を予防する」化学と生物. 2020;58(11):621-627. 公益社団法人日本農芸化学会. https://katosei.jsbba.or.jp/view_html.php?aid=1367

¹¹ Unno K, Pervin M, Taguchi K, Konishi T, Nakamura Y. "Green Tea Catechins Trigger Immediate-Early Genes in the Hippocampus and Prevent Cognitive Decline and Lifespan Shortening." Molecules. 2020;25(7):1484. https://doi.org/10.3390/molecules25071484

¹² お茶百科.「お茶のおいしさを決める水と温度」. 公益社団法人日本茶業中央会. https://www.ocha.tv/how_to_brew/water_and_temperature/

¹³ 京都府消費生活科学センター.「ポリフェノール含有食品比較テスト結果」. 平成11年8月-平成12年2月. https://www.tokyo-cha.or.jp/entry2006/20060405_1254.html

¹⁴ 北海道消費者協会.「緑茶(ペットボトル)のカテキン類含有量」商品テスト結果. https://www.do-syouhi-c.jp/test/kira124ryokutyanokatekin.pdf