Caffeine and Dementia Risk: What the 43-Year JAMA Cohort Study Found

For decades, the question of whether coffee is good or bad for the brain has produced shifting, conflicting headlines. A study published February 9, 2026 in JAMA — led by investigators at Mass General Brigham, Harvard T.H. Chan School of Public Health, and the Broad Institute — adds the largest and longest-running piece of evidence to date. The headline finding: men and women who drank 2-3 cups of caffeinated coffee daily had an 18% lower risk of dementia over up to 43 years of follow-up, and the protective association was strongest in the age 75-and-younger subgroup, where the reduction reached 35%.

This article walks through what the study actually measured, where the dose-response curve seems to level off, why the decaf result matters as much as the coffee result, and how to think about your own intake using the Caffeine Calculator, Caffeine Dose Calculator, and Caffeine Sleep Optimizer.

What the Study Did

The investigators combined data from two of the longest-running prospective cohorts in U.S. epidemiology: the Nurses’ Health Study (NHS) and the Health Professionals Follow-Up Study (HPFS). The pooled analysis included 131,821 participants who were typically enrolled in their early 40s and followed for up to 43 years. Over the follow-up window, 11,033 participants — about 8% — developed dementia.

Coffee and tea intake were captured by repeated food-frequency questionnaires, which allowed researchers to estimate average daily caffeine exposure across decades, not just at a single snapshot. The analysis adjusted for age, sex, education, smoking, alcohol, body mass index, physical activity, hypertension, diabetes, and family history of dementia, among other variables.

The Headline Numbers

• 2-3 cups of caffeinated coffee per day was associated with an 18% lower risk of dementia compared with little or no caffeinated coffee.
• In participants age 75 and younger, the same intake level was associated with a 35% lower dementia risk and 250-300 mg of caffeine per day appeared to be the sweet spot.
• 1-2 cups of caffeinated tea per day showed a similar but somewhat smaller protective association.
• Decaffeinated coffee did not show a protective association. In fact, higher decaf intake was associated with faster memory decline.

The decaf finding is the most useful piece of information for interpreting the rest. If coffee itself — independent of caffeine — were protective, decaf would show some of the same signal. It does not. The plausible read is that the caffeine molecule, not the coffee matrix as a whole, is doing the work. The investigators also note a likely reverse-causation explanation for the decaf result: many people switch from caffeinated to decaffeinated coffee after developing sleep problems, high blood pressure, or cardiac arrhythmias — conditions that themselves carry elevated dementia risk. So heavier decaf drinkers may, on average, be a less metabolically healthy group at baseline.

The Dose-Response Curve

The protective association did not increase indefinitely with higher caffeine intake. Above approximately 300 mg of caffeine per day (roughly three 8-oz cups of brewed coffee), the curve flattened. Higher intakes were not associated with progressively lower dementia risk, and at very high levels the association became harder to interpret because that group is small and tends to differ from average drinkers on other behaviors.

The U.S. Food and Drug Administration’s general guidance for healthy adults is up to 400 mg of caffeine per day. The dose-response curve in this study suggests there is no meaningful brain-aging upside to pushing toward that upper bound, and the trade-offs against sleep, blood pressure, anxiety, and gastrointestinal tolerance get worse as you go higher. The Caffeine Calculator totals your daily intake across coffee, tea, energy drinks, pre-workout, and chocolate so you can see where you actually sit relative to that 250-300 mg sweet spot rather than the 400 mg ceiling.

What This Study Is — and What It Is Not

This is an observational cohort study, not a randomized trial. It cannot establish causation. What it can do — and does well, because of the size and 43-year duration — is show a stable, dose-responsive association that survives adjustment for the most common confounders. The fact that the protective signal appears specifically with caffeinated coffee and tea, and not with decaf, is mechanistically consistent with prior randomized work on caffeine, adenosine receptors, and neuroinflammation.

Some limitations the authors discuss:

• Participants were predominantly white health professionals. Generalization to other populations should be done cautiously.
• Coffee preparation methods (drip, espresso, French press, cold brew) and bean roast level were not separately tracked, and these affect caffeine content per cup substantially.
• The dementia diagnoses were self-reported with medical-record validation in a subset, not adjudicated for every case.
• The 35% reduction was specific to a subgroup analysis (age ≤ 75). Subgroup findings warrant follow-up confirmation in independent cohorts before being treated as definitive.

Caffeine Mechanisms That Are Plausibly Brain-Protective

Why might caffeine be associated with lower dementia risk? Several mechanisms are well established in the laboratory literature, though their relative contribution in humans is still uncertain:

• Adenosine A2A receptor antagonism. Caffeine blocks adenosine signaling at A2A receptors, which are implicated in neuroinflammation and tau pathology in preclinical models of Alzheimer’s disease.
• Reduced beta-amyloid accumulation in animal models. Long-term caffeine in mouse models of Alzheimer’s disease has been associated with lower amyloid plaque burden, though the human evidence is far more limited.
• Antioxidant and anti-inflammatory effects. Coffee — caffeinated or not — contains polyphenols and chlorogenic acids. The fact that decaf does not share the dementia-risk signal in this study suggests these compounds are not the dominant mechanism, even if they have other plausible benefits.
• Cerebrovascular effects. Modest, dose-dependent improvements in cerebral blood flow and endothelial function have been reported with regular caffeine intake.

How to Read This Against Your Own Habits

The practical takeaway from this study is narrower than most headlines suggest. It does not say that coffee prevents dementia. It does not say that people who don’t currently drink coffee should start. What it suggests is that for adults already consuming caffeinated coffee or tea, the 2-3 cups/day range (roughly 250-300 mg of caffeine total) sits in a window associated with lower long-term dementia risk in large U.S. cohorts — and that the upper end of legal caffeine intake offers no additional cognitive benefit while raising the cost on sleep and cardiovascular tolerability.

A few personal-context factors are worth considering before changing your intake:

• Sleep impact: The same caffeine that may benefit long-term brain aging will degrade tonight’s sleep architecture if consumed too late. Sleep quality itself is one of the strongest modifiable risk factors for dementia. Use the Caffeine Sleep Optimizer to see when your last cup drops below the sleep-disruption threshold.
• Cardiovascular conditions: Hypertension, arrhythmias, and certain heart-rhythm disorders interact with caffeine. Anyone with a relevant condition should discuss intake with their clinician before treating this study as license to add cups.
• Pregnancy: ACOG guidance puts the ceiling at 200 mg of caffeine per day during pregnancy, well below the 250-300 mg sweet spot in this study. This study’s findings should not be read as a reason to exceed the pregnancy threshold.
• Genetic metabolism: CYP1A2 polymorphisms substantially change how quickly caffeine clears your system. Slow metabolizers may experience sleep and blood-pressure costs at lower doses than fast metabolizers.
• Anxiety, GERD, migraine: All can be aggravated by caffeine. The right intake for you balances long-term brain-aging signals against current quality of life.

What About People Who Don’t Drink Coffee at All?

The study does not provide evidence that starting coffee de novo will reduce your dementia risk. Lifelong non-drinkers and people who cannot tolerate caffeine have plenty of other evidence-based dementia-risk levers: cardiovascular fitness, hearing protection, sleep regularity, social engagement, midlife blood pressure control, and management of type 2 diabetes are all strongly supported. Tea, which showed a smaller but similar protective signal in this analysis, is a reasonable alternative for anyone who tolerates it but finds coffee too intense.

Using the Caffeine Calculators Together

The three caffeine tools on this site work together to give you a complete picture of your intake:

• The Caffeine Calculator tallies your total daily caffeine across coffee, tea, energy drinks, pre-workout, soft drinks, and chocolate. Use this to see whether you sit in the 250-300 mg JAMA sweet spot, below it, or above it.
• The Caffeine Dose Calculator applies the 3-6 mg/kg pre-exercise research range if you use caffeine specifically for training performance. This is a different question than long-term brain aging — it’s about acute ergogenic dose.
• The Caffeine Sleep Optimizer uses a 5-hour half-life decay model to estimate residual caffeine at bedtime and recommend a personal cutoff time. Long-term brain benefits don’t survive a chronically wrecked sleep schedule.

Together, these answer the three practical questions the JAMA study raises: How much caffeine am I getting? Is that the right amount for what I’m trying to do? And am I timing it in a way that doesn’t cost me sleep tonight?

Bottom Line

A 43-year prospective cohort of 131,821 U.S. health professionals reported that 2-3 cups of caffeinated coffee per day (about 250-300 mg of caffeine) was associated with an 18% lower risk of dementia overall, and a 35% lower risk in the age 75-and-younger subgroup. The dose-response leveled off above 300 mg, decaf did not show the same association, and these are correlations, not proof of cause. For people already drinking coffee or tea, this is the strongest evidence to date that staying in the moderate range — not the maximum — sits in a brain-friendly window. Talk to your clinician before changing your intake if you have a cardiovascular, sleep, anxiety, or pregnancy-related reason to be cautious about caffeine.