Adults who drink two to three cups of coffee and two to three cups of tea each day show a lower rate of dementia than people who consume neither beverage, according to a large prospective analysis of UK Biobank data. The finding, drawn from a cohort of more than 365,000 participants, has gained fresh attention after Harvard reporting in February 2026 highlighted the dose range where the association was strongest. The result also comes with a warning: intake above six cups of coffee per day was tied to smaller brain volumes and higher dementia risk, suggesting the relationship between caffeine and brain health is not a simple “more is better” story.
Why moderate coffee and tea intake draws scientific attention now
Dementia cases are rising sharply as populations age, and few modifiable risk factors have shown consistent associations with lower incidence. Diet is one of the few levers people can actually adjust, which is why a large UK Biobank cohort study in PLOS Medicine attracted wide interest. That analysis tracked participants over a median follow-up period and found that a combined daily habit of two to three cups of coffee plus two to three cups of tea was associated with lower dementia risk compared with drinking none of either beverage. The study reported hazard ratios and confidence intervals for multiple intake categories, and the moderate range consistently stood out.
One reason the combined intake pattern matters is that coffee and tea deliver different bioactive compounds alongside caffeine. Coffee is rich in chlorogenic acids and diterpenes; tea supplies catechins and other flavonoids with documented anti-inflammatory and vascular effects. A plausible biological explanation is that the vasoconstrictive tendency of caffeine, which can reduce cerebral blood flow at high doses, may be partially offset by the vasodilatory properties of tea flavonoids. If that mechanism holds, the net cerebrovascular benefit of drinking both beverages in moderate amounts would differ from what total caffeine dose alone would predict. The UK Biobank data are consistent with that idea, though the study design cannot prove causation.
UK Biobank and Harvard findings on the two-to-three-cup range
The primary evidence comes from a prospective cohort that assessed self-reported coffee and tea consumption at enrollment and then tracked incident dementia, stroke, and poststroke dementia through linked hospital and death records. Methodological details, including diagnostic coding and linkage procedures, align with standard cohort practices described in resources hosted by the National Library of Medicine, which catalogues similar large-scale epidemiologic datasets.
Separate UK Biobank catalogue entries confirm related analyses. One examined coffee intake stratified by genetic risk using polygenic risk scores and found the association held across instant, ground, and decaffeinated coffee types. That consistency across coffee preparations suggests the protective signal is not driven by a single compound unique to one brewing method. Another analysis focused on the upper end of the consumption spectrum. Participants drinking more than six cups of coffee per day showed non-linear risk patterns, including smaller total brain volumes on MRI and elevated dementia risk. That finding is significant because it establishes a ceiling: the apparent benefit seen at two to three cups does not scale upward indefinitely. Very high intake appears to reverse the association.
Harvard reporting published in February 2026 described the most pronounced benefit as appearing in the moderate range. The Harvard Gazette framed the results around study duration, cohorts used, and the specific dose window, while Harvard Health Publishing offered practical context about what “cups” and caffeine levels mean for everyday drinkers and flagged the standard caution that observational associations do not equal causation. Together, these summaries helped translate relatively technical hazard ratios into everyday guidance without overstating what the data can prove.
Open questions about confounding, biology, and long-term follow-up
Several gaps in the evidence prevent a clean causal claim. The UK Biobank cohort is predominantly white and British, which limits how broadly the results can be applied to other populations. Consumption was self-reported at a single time point, and individual caffeine metabolism varies widely based on genetics, medications, and liver function. No biomarker-based caffeine measurements were paired with dementia outcomes in the published analyses, so the actual dose reaching the brain remains uncertain.
Residual confounding is another unresolved issue. People who drink moderate amounts of coffee and tea may differ from non-drinkers in ways that independently affect dementia risk, including sleep quality, socioeconomic status, physical activity, and social engagement. The published analyses adjusted for many lifestyle factors, but no observational study can fully eliminate confounding. Some of these concerns are echoed in researcher profiles and discussion threads linked through My NCBI, where authors organize their publication lists and highlight methodological caveats.
Long-term MRI data paired specifically with tea-only intake categories have been referenced in citation trails but have not been released in the main dataset. That gap matters because it leaves the tea side of the combined-intake finding less well characterized than the coffee side. Without those imaging data, the hypothesis that tea flavonoids contribute a distinct vascular benefit remains biologically plausible but not directly tested in the same rigorous way as high-coffee consumption.
Biological mechanisms also remain partly speculative. Caffeine is known to block adenosine receptors, which can acutely increase alertness but may also influence cerebral blood flow and neuronal excitability. Polyphenols in both coffee and tea have antioxidant and anti-inflammatory properties, and some animal studies suggest they may reduce amyloid deposition or improve endothelial function. However, translating these mechanistic hints into human dementia outcomes requires careful longitudinal work, ideally with repeated imaging and biomarker measurements, not just baseline self-reports.
What the evidence means for everyday drinkers
For individuals trying to interpret these findings, the main takeaway is one of moderation and context rather than a prescription. The UK Biobank data and subsequent Harvard coverage suggest that a daily pattern of roughly two to three cups of coffee and a similar amount of tea is associated with lower dementia risk compared with no intake, while very high coffee consumption may be harmful for brain structure and cognition. That pattern is consistent with a broader literature in which many dietary exposures show U-shaped or J-shaped relationships with chronic disease outcomes.
At the same time, coffee and tea are not risk-free. Caffeine can worsen anxiety, disrupt sleep, and interact with medications. People with certain heart rhythm disorders, uncontrolled hypertension, or pregnancy-related restrictions may need to limit intake well below the ranges highlighted in these analyses. Sugar and cream added to beverages also carry metabolic consequences that are not captured when studies categorize drinks simply as “coffee” or “tea.”
Clinicians and public health practitioners therefore tend to frame the evidence as supportive of moderate coffee and tea consumption within an overall healthy lifestyle, rather than as a stand-alone dementia prevention strategy. Regular physical activity, blood pressure control, smoking cessation, and social engagement all have stronger and more consistently demonstrated links to brain health. Coffee and tea may fit into that picture as potentially beneficial, but not essential, components.
Where research goes next
Future work is likely to focus on several fronts: more diverse cohorts to test whether the associations hold across ethnic and cultural groups; repeated dietary assessments to capture changes in drinking habits over time; and integration of MRI, blood biomarkers, and genetic data to clarify mechanisms. Randomized trials of long-term coffee or tea interventions with dementia as an endpoint are logistically challenging, but shorter studies using cognitive testing and imaging could help bridge the gap between basic biology and population-level observations.
For now, the emerging consensus from UK Biobank analyses and institutional summaries is cautious but encouraging. Moderate daily coffee and tea consumption appears compatible with, and possibly supportive of, healthy brain aging, while very high coffee intake may carry risks that outweigh any potential benefit. As with many areas of nutrition science, the message is less about a miracle beverage and more about balance, individual tolerance, and the cumulative impact of everyday choices over decades.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
