Is Coffee Good for You? The Evidence Says “2–4 Cups a Day, Filtered, Before 2 p.m.” [2026 Edition]

[Level 2 (Moderate–High)] [Food & Nutrition] [Series: Health for Serious Middle-Agers]

⚠️ Before You Read This Article

This article provides educational information based on peer-reviewed studies on coffee and caffeine (large cohorts, umbrella reviews, and RCTs). It is not a substitute for medical advice. If you have severe hypertension (160/100 mmHg or higher), an arrhythmia, insomnia, or are pregnant, please use the condition-specific guidance in this article as a starting point and consult your physician.

Key Points (3-Line Summary)

• Large cohorts and umbrella reviews consistently agree that 2–4 cups a day is associated with the lowest risk of all-cause mortality, cardiovascular disease, type 2 diabetes, and liver disease (JPHC: HR 0.76 at 3–4 cups). Beyond 6 cups there is no added benefit, and risk rises in a J-shaped curve.
• Blood pressure rises acutely, but habitual drinkers develop tolerance, and coffee does not increase atrial fibrillation (confirmed by RCT). The exception is severe hypertension: in a Japanese study, 2 or more cups was associated with cardiovascular mortality HR 2.05.
• The effect on sleep is certain. 400 mg of caffeine taken 6 hours before bedtime cuts sleep by 1 hour. “Coffee until 2 p.m., decaf after that” is the practical answer.

1. Mortality, Cardiovascular Disease, Diabetes, and Liver Disease

Few foods have been studied in as many large cohorts as coffee, and the conclusions are broadly consistent.

• JPHC (Japan, approx. 90,000 participants): 3–4 cups a day was associated with all-cause mortality HR 0.76. Deaths from heart disease, cerebrovascular disease, and respiratory disease were lower.
• NIH-AARP (US, approx. 400,000) and UK Biobank (approx. 500,000): both showed lower all-cause mortality in coffee drinkers. The dose-response curve bottoms out at 2–4 cups and plateaus beyond that.
• Umbrella review (Poole et al., BMJ 2017): pooled 201 meta-analyses. At 3–4 cups, all-cause mortality RR 0.83 and cardiovascular mortality RR 0.81, with lower rates of type 2 diabetes, liver cirrhosis, and liver cancer. The only consistently demonstrated harms were high intake during pregnancy and fractures in women.
• More than 6 cups: Zhou et al. (2019, UK Biobank) found a J-shaped association with cardiovascular disease HR 1.22.

The limitation is the bias inherent in observational research. Sick people cut back on coffee (reverse causation), smoking is a confounder, and UK Biobank participants tend to be health-conscious. Some Mendelian randomization studies (which use genetic variants as a stand-in for randomization) have failed to reproduce coffee’s protective effect, so we cannot go as far as to call it “causal.” Even so, it is clear that signals pointing toward harm are extremely scarce.

2. Blood Pressure and Arrhythmia: Yes, It “Goes Up,” But…

Caffeine acutely raises blood pressure by about +8/+6 mmHg. In habitual drinkers, however, tolerance develops, and the long-term risk of developing hypertension does not rise and, if anything, tends to fall (Mesas et al., meta-analysis 2011).

For arrhythmias, we have RCTs. In CRAVE (NEJM 2023), premature atrial contractions did not increase on coffee-drinking days, and in DECAF (JAMA 2025, patients after cardioversion for atrial fibrillation), the coffee group actually had less recurrence of atrial fibrillation (HR 0.61). The conventional wisdom that “coffee triggers atrial fibrillation” is not supported by current evidence.

There is, however, one important exception. In the Japanese JACC study (Teramoto et al., J Am Heart Assoc 2023), people with severe hypertension of 160/100 mmHg or higher who drank 2 or more cups a day had cardiovascular mortality HR 2.05. There was no association in mild-to-moderate hypertension, and none for green tea. If you have severe hypertension, the safe course is to keep it to 1 cup or less, or switch to green tea.

3. Sleep: The Biggest Pitfall

Caffeine has a half-life of about 5 hours, with large individual variation (the CYP1A2 gene).

• Drake et al. (J Clin Sleep Med 2013): when 400 mg was taken 0, 3, or 6 hours before bedtime, even the 6-hour dose cut total sleep time by about 1 hour, and participants were not aware of it.
• Gardiner et al. (Sleep Med Rev 2023, meta-analysis): caffeine reduced total sleep time by 45 minutes, lengthened sleep onset latency by 9 minutes, and reduced sleep efficiency by 7%. To avoid these effects, a cup of coffee (about 107 mg) should be consumed no later than 8.8 hours before bedtime, and a pre-workout-level dose (217 mg) no later than 13.2 hours before.

This issue ties directly into evidage’s #5 (sleep regularity) and #9 (light hygiene). A Japanese cup (150 mL) contains about 90 mg, so “decaf after 2 p.m.” is the most realistic rule.

4. Bone, Stomach, and Pregnancy

• Bone: bone density falls by 2–4%, but fractures increase only slightly in women (RR 1.14) and show no association in men. With adequate calcium intake, the real-world harm is small.
• Stomach: an endoscopic study in Japanese subjects found no association with gastric ulcer or reflux esophagitis.
• Pregnancy: under 200 mg/day (2 cups of coffee) is the upper limit set by national guidelines.

5. Brewing Method Matters: Filtered Is Safest

Cafestol, found in the oils of coffee beans, raises LDL cholesterol. Boiled coffee and French press deliver 4–12 mg per cup, while paper-filtered coffee contains under 0.1 mg. In Tverdal et al. (Eur J Prev Cardiol 2020, 510,000 Norwegians), filtered-coffee drinkers had the lowest mortality, while unfiltered coffee was associated with higher cardiovascular mortality in men.

As for decaf and instant coffee, Chieng et al. (Eur J Prev Cardiol 2022, UK Biobank) found both associated with lower mortality, suggesting that the benefit comes mainly from polyphenols and other compounds rather than caffeine. If you are sensitive to caffeine, decaf is perfectly fine.

6. Japanese Data and Green Tea

The Fukuoka Diabetes Registry reported all-cause mortality HR 0.37 for the combination of 2 or more cups of coffee plus 4 or more cups of green tea. Green tea alone, in a pooled analysis of 8 Japanese cohorts, was associated with all-cause mortality HR of about 0.87 at 5 or more cups (figure to be verified). Coffee and green tea are not competitors; what the Japanese data show is that combining them is beneficial.

The safe upper limit is 400 mg/day per EFSA (healthy adults). At about 90 mg per Japanese cup, 4 cups comes to 360 mg.

7. Conditional Conclusions

There is no need to drink coffee “for your health,” but if you enjoy it, there is no reason to stop, as long as you stick to 2–4 cups, filtered, before 2 p.m.

8. Frequently Asked Questions

Q. Does the same apply to canned coffee and caffe latte? A. Sweetened versions add a separate set of problems related to UPF and sugar (see #4 and #7). Stick to black or lightly sweetened.

Q. Is the caffeine in energy drinks treated the same way? A. Even if the caffeine content is similar, the burden of sugar and additives is substantial, so they cannot be treated as equivalent (see our existing article “Energy Drinks”).

Q. What if I can only drink coffee in the evening? A. Switch to decaf. You still get the benefits of the polyphenols with decaf.

References and Sources

• Poole R, et al. (2017). Coffee consumption and health: umbrella review of meta-analyses. BMJ 359:j5024.
• Saito E, et al. (2015). Association of coffee intake with total and cause-specific mortality in a Japanese population (JPHC). Am J Clin Nutr 101:1029-1037.
• Zhou A, Hyppönen E. (2019). Long-term coffee consumption, caffeine metabolism genetics, and risk of cardiovascular disease. Am J Clin Nutr 109:509-516.
• Mesas AE, et al. (2011). The effect of coffee on blood pressure and cardiovascular disease in hypertensive individuals. Am J Clin Nutr 94:1113-1126.
• Marcus GM, et al. (2023). Acute effects of coffee consumption on health among ambulatory adults (CRAVE). N Engl J Med 388:1092-1100.
• DECAF trial. (2025). Coffee consumption and atrial fibrillation recurrence. JAMA. doi:10.1001/jama.2025.21056. PMID: 41206802
• Teramoto M, et al. (2023). Coffee and green tea consumption and cardiovascular disease mortality among people with and without hypertension (JACC). J Am Heart Assoc 12:e026477.
• Drake C, et al. (2013). Caffeine effects on sleep taken 0, 3, or 6 hours before going to bed. J Clin Sleep Med 9:1195-1200.
• Gardiner C, et al. (2023). The effect of caffeine on subsequent sleep: a systematic review and meta-analysis. Sleep Med Rev 69:101764.
• Tverdal A, et al. (2020). Coffee consumption and mortality from all causes, cardiovascular disease and cancer: does the brewing method matter? Eur J Prev Cardiol 27:1986-1993.
• Chieng D, et al. (2022). The impact of coffee subtypes on incident cardiovascular disease, arrhythmias, and mortality. Eur J Prev Cardiol 29:2240-2249.
• EFSA (2015). Scientific Opinion on the safety of caffeine.

Evidence current as of October 2026. We recommend verifying figures such as the HR from the pooled analysis of 8 green tea cohorts against the original publications.