- In huge studies, coffee drinkers die at lower rates: 2 to 3 cups a day was linked to a 12% lower risk of death in UK Biobank (498,134 adults), and the link held for decaf and slow caffeine metabolisers
- That's an association, not proof. Before adjusting for smoking, coffee drinkers looked worse, and a genetic study found "limited evidence" that coffee changes heart risk
- Heart rhythm trials are reassuring: people with atrial fibrillation told to drink coffee had less recurrence (47% vs 64%). But coffee days brought more extra beats from the lower chambers and 36 minutes less sleep
- Filter it. French press and boiled coffee carry oils that raise cholesterol; a paper filter removes most of them
- Limits: the FDA cites 400 mg of caffeine a day for most adults; under 200 mg in pregnancy. Avoid it late in the day
Headlines say coffee drinkers live longer. The big studies are consistent, but they can't show that coffee is the cause, and the randomised trials measure short-term effects, not lifespan. Here's what each type of evidence shows and how to drink it sensibly. It isn't medical advice: if you have a heart condition, high blood pressure or trouble sleeping, talk to your doctor about your intake.
What the Big Cohorts Found
Evidence: Consistent associations — observational onlyNIH-AARP: 400,000 Americans
A 2012 New England Journal of Medicine study followed 229,119 men and 173,141 women aged 50 to 71 from 1995 to 2008; 33,731 men and 18,784 women died. After adjusting for smoking and other factors, men drinking 4 or 5 cups a day had a hazard ratio for death of 0.88 compared with non-drinkers (about 12% lower), and women 0.84. Coffee was linked to fewer deaths from heart disease, respiratory disease, stroke, injuries and accidents, diabetes and infections, but not cancer. The authors wrote: "Whether this was a causal or associational finding cannot be determined from our data" (Freedman, NEJM 2012).
UK Biobank: half a million people, and their genes
In 2018, Loftfield and colleagues analysed 498,134 UK Biobank participants aged 38 to 73, of whom 14,225 died over 10 years. Risk of death was lower than in non-drinkers at every level from 1 cup a day to 8 or more (Loftfield, JAMA Intern Med 2018).
Risk of death vs non-drinkers, by cups a day
Loftfield et al., JAMA Internal Medicine 2018. Adjusted hazard ratios for death from any cause, vs non-drinkers (1.00). *Not statistically significant (95% CI 0.88 to 1.01). Observational data: associations, not proof of cause.
The twist: some people clear caffeine slowly because of gene variants, so heavy coffee might be riskier for them. The researchers found "no evidence of effect modification across strata of caffeine metabolism score". The pattern looked similar in slow and fast metabolisers, and held for instant, ground and decaffeinated coffee. The decaf result hints that if coffee helps, caffeine may not be the reason. The authors cautioned that their results "are based on observational data and should be interpreted with caution."
What about sugar?
A 2022 Annals of Internal Medicine study followed 171,616 UK Biobank participants free of heart disease and cancer; 3,177 died over a median of 7 years. Compared with non-drinkers, more than 1.5 and up to 3.5 cups a day had hazard ratios of 0.84 to 0.71 for unsweetened coffee and 0.69 to 0.72 for sugar-sweetened. Sugar-sweetened coffee wasn't linked to lower risk at the lowest or highest intakes. Results for artificially sweetened coffee were "less consistent" (Liu, Ann Intern Med 2022). But people who sweetened their coffee added only about 1 teaspoon of sugar per cup on average (American College of Physicians), a long way from a large flavoured coffee-shop drink.
The umbrella review: 201 meta-analyses
A 2017 review in The BMJ examined 201 meta-analyses of observational studies covering 67 health outcomes, plus 17 meta-analyses of trials. Coffee "was more often associated with benefit than harm." For death from any cause, the relative risk was 0.83 at three cups a day versus none, and the review found the biggest benefits generally at three to four cups. Two exceptions stood out: in pregnancy, high intake was linked to low birth weight (odds ratio 1.31), preterm birth and pregnancy loss (1.46), and coffee was linked to fracture in women but not men. Coffee "seems generally safe within usual levels of intake," the authors concluded, but most evidence was "low or very low quality" (Poole, BMJ 2017).
Why an Association Isn't Proof
Evidence: Genetic studies argue for cautionCoffee drinkers differ from non-drinkers in ways that are hard to measure. In the NIH-AARP data, adjusted only for age, "the risk of death was increased among coffee drinkers", but they were also more likely to smoke. Only after adjusting for smoking did coffee look protective, and the BMJ review found that apparent harms mostly disappeared once smoking was accounted for. If one factor can flip the result, others could tilt it. People who quit coffee because they feel unwell also land in the "non-drinker" group. And both big cohorts measured coffee only once, at the start.
Mendelian randomisation gets around some of this by using gene variants that nudge people to drink more or less coffee. Genes are set at conception, so they are less tangled up with lifestyle, although some coffee-related genes also influence smoking, which the study adjusted for. A 2021 study in Nutrients used 12 such variants, with data from UK Biobank (up to 35,979 cases) and Finland's FinnGen (up to 17,325 cases). Genetically predicted coffee drinking "was not associated with any of the 15 studied cardiovascular outcomes"; only after adjusting for body weight and smoking did a suggestive link with bleeding in the brain appear, in FinnGen. The authors found "limited evidence that coffee consumption affects the risk of developing cardiovascular disease" and suggested earlier observational findings "may have been confounded" (Yuan, Nutrients 2021).
Coffee and Your Heart Rhythm
Evidence: Randomised trials — short-termCRAFT: coffee days vs no-coffee days
In the CRAFT trial, led by Gregory Marcus at the University of California, San Francisco, 100 healthy adults wore a continuous ECG and devices tracking steps, sleep and blood sugar for 14 days. By text message, they were "randomly assigned to consume caffeinated coffee or to avoid caffeine during 2-day periods" (Marcus, NEJM 2023). Coffee days compared with caffeine-free days:
- Early beats from the upper chambers: 58 vs 53 a day, not a statistically significant difference.
- Early beats from the lower chambers: 154 vs 102 a day, about 51% more.
- Steps: about 1,000 more a day (10,646 vs 9,665).
- Sleep: 36 minutes less a night (397 vs 432 minutes).
- Blood sugar: no meaningful difference.
The authors cautioned that the volunteers were "relatively young and healthy," so the results may not apply to people who already have frequent extra beats.
DECAF: coffee in people with atrial fibrillation
The DECAF trial, presented in November 2025 and published in JAMA, enrolled 200 adults with persistent atrial fibrillation (AF) from 5 hospitals in the US, Canada and Australia, all current or recent coffee drinkers. After cardioversion to reset the heart's rhythm, they were randomised to drink at least 1 cup of caffeinated coffee a day or to avoid all coffee, decaf included, and other caffeine for 6 months. AF or atrial flutter came back in 47% of the coffee group and 64% of the abstinence group, a 39% lower hazard of recurrence, with no significant difference in adverse events (Wong, JAMA 2026). It was open-label, and it only enrolled people who already drank coffee (ACC).
Together: moderate coffee doesn't seem to trigger atrial rhythm problems for most people. But it does cause more extra beats from the lower chambers and less sleep. If you get palpitations after coffee, raise it with your doctor.
Filtered vs Unfiltered Coffee
Evidence: Strong for cholesterol — moderate for mortalityCoffee beans contain oils called diterpenes, mainly cafestol and kahweol, that raise cholesterol. Research cited by Norway's Tromsø team found that every 10 mg of cafestol, plus a similar amount of kahweol, raised total cholesterol by 0.13 mmol/L. A paper filter holds back most of them: one analysis they cite found 232 mg of cafestol per litre in boiled coffee and 5 mg in filtered (Svatun, Open Heart 2022). In a 2012 meta-analysis of 12 trials with 1,017 people, coffee raised total cholesterol by 8.1 mg/dL and LDL by 5.4 mg/dL on average, more "in trials using unfiltered coffee" (Cai, Eur J Clin Nutr 2012).
The Tromsø Study (21,083 adults aged 40 and over) found six or more cups a day of boiled or plunger (French press) coffee linked to total cholesterol 0.23 to 0.30 mmol/L higher, and three to five espressos to 0.09 to 0.16 mmol/L higher (Svatun 2022). A 2020 study followed 508,747 Norwegians aged 20 to 79 for about 20 years. Compared with men who drank no coffee, men drinking filtered coffee had a hazard ratio for death of 0.85, and unfiltered 0.96, which wasn't significant; for women, 0.85 and 0.91. "Unfiltered brew was associated with higher mortality than filtered brew," the authors concluded (Tverdal, Eur J Prev Cardiol 2020).
How you brew it: cholesterol and mortality
| Brew | Cafestol per litre | Total cholesterol (Tromsø) | Risk of death vs no coffee |
|---|---|---|---|
| Paper-filtered | 5 mg | ≥6 cups/day: +0.11 mmol/L in women; no link in men | HR 0.85 men, 0.85 women |
| Boiled or plunger (French press) | 232 mg boiled | ≥6 cups/day: +0.30 women, +0.23 men | HR 0.96 men not significant 0.91 women "unfiltered" brew |
| Espresso | ~54 mg espresso machines | 3–5 cups/day: +0.09 women, +0.16 men | — not reported separately |
| Capsule, pod, vending | 10–43 mg | — not reported separately | — |
Cafestol and cholesterol: Svatun et al., Open Heart 2022 (Tromsø Study, 21,083 adults, cross-sectional; cafestol figures from earlier studies it cites). Mortality: Tverdal et al., Eur J Prev Cardiol 2020 (508,747 adults). Both observational. 1 mmol/L of total cholesterol is about 39 mg/dL.
Blood pressure
The FDA lists high blood pressure among the symptoms of too much caffeine. In a Japanese study of more than 18,600 adults aged 40 to 79 followed for nearly 19 years, people with blood pressure of 160/100 mm Hg or higher who drank two or more cups a day had twice the risk of cardiovascular death of non-drinkers. One cup a day wasn't linked to higher risk at any blood pressure level. Intake was self-reported and blood pressure measured once (Teramoto, JAHA 2022).
How Much Caffeine Is Too Much?
Evidence: Official guidance"For most adults, the FDA has cited 400 milligrams a day," about two to three 12-ounce cups of coffee, "as an amount not generally associated with negative effects." The FDA notes "wide variation" in how sensitive people are to caffeine and how fast they clear it. Signs of too much include a faster heart rate, palpitations, anxiety, jitters and sleep problems (FDA). The European Food Safety Authority (EFSA) also uses 400 mg a day for healthy adults, and up to 200 mg in one dose (EFSA). In pregnancy, the American College of Obstetricians and Gynecologists says moderate caffeine consumption, "less than 200 mg per day," "does not appear to be a major contributing factor in miscarriage or preterm birth" (ACOG).
Caffeine per drink, and the limits
| Drink | Serving | Caffeine |
|---|---|---|
| Brewed coffee regular, non-specialty | 12 fl oz | 113–247 mg |
| Decaf coffee | 8 fl oz | 2–15 mg |
| Black tea | 12 fl oz | 71 mg |
| Green tea | 12 fl oz | 37 mg |
| Caffeinated soft drink | 12 fl oz | 23–83 mg |
| Energy drink | 12 fl oz | 41–246 mg |
| Limit, most healthy adults | per day | 400 mg FDA, EFSA |
| Limit in pregnancy | per day | under 200 mg ACOG; EFSA 200 mg |
FDA, "Spilling the Beans" (current as of 28 August 2024); EFSA (2015); ACOG Committee Opinion 462 (reaffirmed 2023). Brewed coffee varies twofold.
Timing and sleep
EFSA puts caffeine's average half-life at about four hours, with a range of two to eight. On that average, about a quarter of a 3 pm coffee's caffeine is still circulating at 11 pm. In a 2013 study, 12 healthy sleepers took 400 mg of caffeine, a large dose, at bedtime, 3 hours before, or 6 hours before. Even at 6 hours, total sleep time fell by more than an hour. The researchers advised avoiding caffeine after 5 pm (American Academy of Sleep Medicine, on Drake 2013). Our sleep and longevity guide covers why that matters.
A Sensible Coffee Habit
No need to start if you don't enjoy it. If you do:
- Stay moderate. The BMJ review found the largest drop in risk at about three cups a day. Café cups can be big and strong, so count caffeine, not cups.
- Filter it, especially if your LDL is high. Paper-filtered drip or pour-over removes most of the cholesterol-raising oils. If you drink French press, boiled coffee or several espressos a day, get your cholesterol checked.
- Keep it plain or lightly sweetened. About a teaspoon of sugar per cup was what UK Biobank studied, not syrups and whipped cream.
- Watch the clock. Have your last caffeinated cup by early afternoon, or switch to decaf.
- Cut back if you get palpitations, anxiety or insomnia, if your blood pressure is very high or uncontrolled, or if you're pregnant or trying to be. Women at high risk of fracture should discuss intake with their doctor.
- Check your medicines. Some medications make people more sensitive to caffeine; ask your pharmacist.
Frequently Asked Questions
Safe for most, not a longevity drug.
Moderate coffee drinkers die at lower rates than non-drinkers, and trials find coffee doesn't seem to trigger atrial fibrillation. But the studies can't separate coffee from the people who drink it, and genetic evidence suggests the heart benefit may be smaller than it looks.
If you enjoy it, drink it filtered, plain or lightly sweetened, and early in the day. Cut back if it gives you palpitations, anxiety or poor sleep, if your blood pressure is very high, or if you're pregnant, and talk to your doctor about what's right for you.
Sources
- Freedman ND, et al. Association of coffee drinking with total and cause-specific mortality. New England Journal of Medicine 2012;366:1891–1904 (doi:10.1056/NEJMoa1112010). Abstract via PubMed 22591295; full text (PMC).
- Loftfield E, Cornelis MC, Caporaso N, Yu K, Sinha R, Freedman N. Association of coffee drinking with mortality by genetic variation in caffeine metabolism: findings from the UK Biobank. JAMA Internal Medicine 2018;178(8):1086–1097 (doi:10.1001/jamainternmed.2018.2425).
- Liu D, et al. Association of sugar-sweetened, artificially sweetened, and unsweetened coffee consumption with all-cause and cause-specific mortality: a large prospective cohort study. Annals of Internal Medicine 2022;175(7):909–917 (doi:10.7326/M21-2977). Summary via UK Biobank; ACP news release.
- Poole R, Kennedy OJ, Roderick P, Fallowfield JA, Hayes PC, Parkes J. Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes. BMJ 2017;359:j5024 (doi:10.1136/bmj.j5024).
- Yuan S, Carter P, Mason AM, Burgess S, Larsson SC. Coffee consumption and cardiovascular diseases: a Mendelian randomization study. Nutrients 2021;13(7):2218 (doi:10.3390/nu13072218).
- Marcus GM, et al. Acute effects of coffee consumption on health among ambulatory adults. New England Journal of Medicine 2023;388(12):1092–1100 (doi:10.1056/NEJMoa2204737).
- Wong CX, Cheung CC, Montenegro G, et al. Caffeinated coffee consumption or abstinence to reduce atrial fibrillation: the DECAF randomized clinical trial. JAMA 2026;335(4):317–325 (doi:10.1001/jama.2025.21056); published online 9 November 2025, presented at AHA Scientific Sessions November 2025. See also American College of Cardiology summary.
- Svatun ÅL, Løchen ML, Thelle DS, Wilsgaard T. Association between espresso coffee and serum total cholesterol: the Tromsø Study 2015–2016. Open Heart 2022;9(1):e001946 (doi:10.1136/openhrt-2021-001946). Abstract via UiT Munin.
- Cai L, Ma D, Zhang Y, Liu Z, Wang P. The effect of coffee consumption on serum lipids: a meta-analysis of randomized controlled trials. European Journal of Clinical Nutrition 2012;66(8):872–877 (doi:10.1038/ejcn.2012.68).
- Tverdal A, Selmer R, Cohen JM, Thelle DS. Coffee consumption and mortality from cardiovascular diseases and total mortality: does the brewing method matter? European Journal of Preventive Cardiology 2020;27(18):1986–1993 (doi:10.1177/2047487320914443).
- Teramoto M, et al. Coffee and green tea consumption and cardiovascular disease mortality among people with and without hypertension. Journal of the American Heart Association 2023;12(2):e026477, published online 21 December 2022 (doi:10.1161/JAHA.122.026477). As reported by the American Heart Association.
- US Food and Drug Administration. Spilling the Beans: How Much Caffeine is Too Much? Content current as of 28 August 2024.
- European Food Safety Authority. EFSA explains risk assessment: Caffeine. 2015.
- American College of Obstetricians and Gynecologists. Committee Opinion No. 462: Moderate caffeine consumption during pregnancy. Obstetrics & Gynecology 2010; reaffirmed 2023.
- Drake C, Roehrs T, Shambroom J, Roth T. Caffeine effects on sleep taken 0, 3, or 6 hours before going to bed. Journal of Clinical Sleep Medicine 2013;9(11):1195–1200 (doi:10.5664/jcsm.3170). As reported by the American Academy of Sleep Medicine.