Does coffee help you live longer?
Short answer
Linked to lower death risk; cause unproven.
Studies that followed large groups for years found coffee drinkers less likely to die from any cause, most clearly at a few cups a day. No trial here tested whether coffee causes this.
This page says what the research shows. It does not say whether you should drink coffee.
Where the evidence stands
Each dot is one on-topic paper. Steps near the top answer the question more directly. Select a dot to see the study.
↑ Stronger evidence
-
Trials in people: death from any cause
Trials that measured death from any cause directly.
0Not done yet. This is the step that would answer the question.
-
Trials in people: other measures
Trials that measured other health or performance measures.
0None among the papers we read.
-
Trials in people: lab tests or safety
Trials that measured lab tests, side effects or safety.
0None among the papers we read.
-
Population studies
Studies that followed or surveyed groups of people. They show links, not causes.
7 -
Blood levels and genes
Studies that link blood levels or genes to health.
0None among the papers we read.
↓ Weaker evidence
- Points to benefit
- Mixed
- No difference
- No direction
- Points to harm
Not on the ladder: 3 reviews of many studies (shown below), 5 opinion or narrative papers. Counts are papers, not people. Some studies appear in more than one paper.
See every studyWhat we found
- A pooled analysis of 40 studies and 3,852,651 people found the lowest risk of death from any cause at 3.5 cups a day. Kim 2019
- In 171 616 UK adults, unsweetened coffee drinkers at every amount had hazard ratios of 0.71 to 0.84 for death from any cause; below 1 means lower risk than non-drinkers. Liu 2022
- In 43 114 US adults, people drinking 1 to 2 cups a day had an estimated life expectancy at age 50 that was 2·02 years longer than non-drinkers'. Yan 2025
- None of these papers is a trial of coffee and death. Reviews say most data are observational and randomised trials are needed to show cause. Poole 2017, Barrea 2023, O'Keefe 2018, Emadi 2025
What the research says
Coffee drinkers died less often from any cause
A pooled analysis of 40 studies found a relative risk of 0.85 at 3.5 cups a day versus none. A US study of 90,317 adults found a hazard ratio of 0.82 at 2-3 cups a day. A UK study found 0.71 to 0.84 for unsweetened coffee.
A ratio below 1 means fewer deaths than among non-drinkers during follow-up. The pooled link held for smokers and non-smokers, and across age and weight. These studies show a link, not that coffee causes longer life.
Studies: Kim 2019, Loftfield 2015, Liu 2022
More cups did not keep lowering risk
The drop in risk was largest at moderate amounts. A review of pooled analyses found the biggest drop at three to four cups a day versus none (relative risk 0.83). The pooled analysis saw no further drop above 3.5 cups. A UK study and a later review describe U-shaped curves.
In US adults, estimated life expectancy at age 50 was 30·06 years for non-drinkers, 32·08 years at 1 to 2 cups a day and 31·45 years above 3 cups.
Studies: Poole 2017, Kim 2019, Liu 2022, Ungvari 2024, Yan 2025
The link also appears in people with illness
Among US adults with type 2 diabetes, the highest coffee intake went with a hazard ratio of 0.74 for death from any cause versus the lowest. In people with an irregular heartbeat, daily coffee went with lower death from any cause.
A review of bowel cancer survivors rated the coffee link limited-suggestive. An older study of 10,064 people with high blood pressure found more caffeine did not mean more deaths over 4 years. A commentary says one study saw the link especially in people with heart or brain blood vessel disease.
Studies: Ma 2023, Iten 2024, Chan 2024, Martin 1988, Mattioli 2025
Studies disagree on sugar; decaf showed the link
In 46,222 US adults, lower death risk was limited to black coffee (hazard ratio 0.86) and coffee low in added sugar and saturated fat. In the UK study, sugar-sweetened coffee also went with lower risk at more than 1.5 to 3.5 drinks a day (0.69, 0.72).
Decaf coffee showed a similar link in the pooled analysis, the bowel cancer review and a US study of 90,317 adults, which also found similar results with coffee additives. The UK study saw U-shaped curves for decaf too.
Studies: Zhou 2025, Liu 2022, Kim 2019, Chan 2024, Loftfield 2015
What reviews of many studies found
Reviews pool or sum up earlier studies. They are not on the ladder, because they repeat studies already counted there.
Pooled review (meta-analysis) · Kim 2019
Pooled 40 studies with 3,852,651 people and 450,256 deaths. Death from any cause was lowest at 3.5 cups a day (relative risk 0.85), with no further drop at higher intake. The link was weaker in the US.
Pooled review (meta-analysis) · Poole 2017
Reviewed 201 pooled analyses of observational research and 17 of trials. Three to four cups a day versus none went with lower death from any cause (relative risk 0.83). It calls for trials to test cause.
Pooled review (meta-analysis) · Chan 2024
Included five trials and 35 observational studies of bowel cancer survivors (30,242 cases). It rated the link between coffee, including decaf, and lower death from any cause as limited-suggestive evidence.
How the research grew
The papers we read span 1988 to 2025. 7 of the 15 on-topic papers came out in 2024 to 2026.
- 1985–89
- 2015–19
- 2020–24
- 2025–26
Each dot is one on-topic paper, colored by what it found. Rings mark key papers:
- 1988. Early study finds caffeine did not raise deaths. Martin 1988
- 2015. Large US study links coffee to lower death. Loftfield 2015
- 2019. Largest pooled analysis here: 40 studies, 3,852,651 people. Kim 2019
- 2022. UK study separates sugar-sweetened from unsweetened coffee. Liu 2022
- 2025. Study turns the coffee link into life expectancy. Yan 2025
Who was studied
Who took part, and how many. There are no trials in people here. The largest population study had 171,616 people.
- Population, blood and gene studies
- 90,317 people · Loftfield 2015
US adults without cancer or heart disease at the start, followed up to 2009 - 46,222 people · Zhou 2025
US adults aged 20 and older in a national health survey, followed about 9 to 11 years - 10,064 people · Martin 1988
Adults with high blood pressure in a treatment and follow-up program, 1973 to 1979 - People in the study, log scale1101001,00010,000100,0001,000,000
- Points to benefit
- Mixed
- No difference
- No direction
- Points to harm
Bars use a log scale: each step is ten times more people. Reviews, animal studies and papers with no count in the abstract are not shown.
Who funds this research
None of the 15 relevant papers is a trial. One review has an author who consulted for the National Coffee Association. No population study shows a coffee company tie; eight papers have no disclosure in PubMed. Stakes did not change the confidence.
- No disclosure in PubMed 5
- No stake declared 5
- Kim 2019No disclosure in PubMed
- Chan 2024No disclosure in PubMed
- Martin 1988No disclosure in PubMed
- Loftfield 2015No disclosure in PubMed
- Liu 2022No disclosure in PubMed
- Poole 2017No stake declared
- Zhou 2025No stake declared
- Iten 2024No stake declared
- Ma 2023No stake declared
- Yan 2025No stake declared
Funding rule. A stake is always shown. Confidence is lowered only when every supporting study comes from a party with a stake. "No disclosure in PubMed" means the PubMed record has no conflict-of-interest statement. It does not mean there is no conflict.
All 15 on-topic papers
Each paper shows what it tested, what it found, and who has a stake. Every paper also has its own page that explains how much it counts and how it fits. See every study as a list.
-
Coffee Consumption and Mortality among United States Adults: A Prospective Cohort Study.
46,222 people · US adults aged 20 and older in a national health survey, followed about 9 to 11 years
TestedCompared people by cups of coffee a day, and by sugar and fat added to the coffee
7074 deaths. Risk of death from any cause was HR 0.84 (0.77, 0.92) at 1 to <2 cups a day and 0.85 (0.77, 0.95) at 3 or more cups. The gain was limited to black coffee and low sugar and fat coffee.
Points to benefitNo stake declaredAbstract onlyFunded by the National Institutes of Health; no conflicts reported.
-
Association between coffee consumption and life expectancy: a prospective cohort study from NHANES 2001-2018.
43,114 people · US adults aged 20 and older
TestedCups of coffee a day, from 0 to over 3, followed a median 8·7 years
6234 deaths. Life expectancy at age 50 was 30·06 years for non-drinkers and 32·08 years for 1 to 2 cups a day, a gain of 2·02 years (1·17, 2·85).
Points to benefitNo stake declaredAbstract onlyNo conflicts of interest reported.
-
Coffee consumption and adverse cardiovascular events in patients with atrial fibrillation.
Adults with an irregular heartbeat (atrial fibrillation) in two Swiss follow-up studies
TestedCompared daily coffee drinkers with people who did not drink it daily, followed a median of 4.7 years
Daily coffee went with a lower risk of major heart events (HR 0.77 (0.66; 0.89)). The authors also report lower all-cause mortality, but give no number in the abstract.
Points to benefitNo stake declaredAbstract onlyNo competing interests declared.
-
Beverage consumption and mortality among adults with type 2 diabetes: prospective cohort study.
15,486 people · US health workers with type 2 diabetes, followed about 18.5 years
TestedCompared highest vs lowest intake of coffee and other drinks
7638 deaths. Highest vs lowest coffee intake had lower death from any cause (pooled hazard ratio 0.74 (0.63 to 0.86)). Sugary drinks went with higher risk (1.20).
Points to benefitNo stake declaredAbstract onlyNIH funded; one author has ties to a walnut group and supplement firms (Metagenics, Standard Process), not coffee makers.
-
Association of Sugar-Sweetened, Artificially Sweetened, and Unsweetened Coffee Consumption With All-Cause and Cause-Specific Mortality : A Large Prospective Cohort Study.
171,616 people · UK Biobank adults, average age 55.6, without heart disease or cancer
TestedUnsweetened, sugar-sweetened and artificially sweetened coffee, from none to more than 4.5 drinks a day, followed a median 7.0 years
3177 deaths. Every amount of unsweetened coffee went with lower death from any cause than no coffee (hazard ratios 0.71 to 0.84). Sugar-sweetened coffee went with lower risk at more than 1.5 to 3.5 drinks a day (0.69, 0.72), but not above 4.5 (1.05).
Points to benefitNo disclosure in PubMedAbstract onlyNo disclosure in PubMed; funders were Chinese science foundations.
-
Association of Coffee Consumption With Overall and Cause-Specific Mortality in a Large US Prospective Cohort Study.
90,317 people · US adults without cancer or heart disease at the start, followed up to 2009
TestedCompared coffee drinkers with nondrinkers by cups a day
8,718 deaths. Risk of death from any cause vs nondrinkers was HR = 0.82 (95% CI: 0.77, 0.88) at 2-3 cups a day and HR = 0.79 (95% CI: 0.72, 0.86) at 4-5 cups. Under 1 cup showed HR = 0.99.
Points to benefitNo disclosure in PubMedAbstract onlyNo disclosure in PubMed; funded by the US National Institutes of Health.
-
Mortality patterns among hypertensives by reported level of caffeine consumption.
10,064 people · Adults with high blood pressure in a treatment and follow-up program, 1973 to 1979
TestedTotal caffeine from coffee, tea and some medicines, measured at the 1-year visit and compared across levels; followed for 4 years
No link was found between more caffeine and more death from any cause or from heart disease over the next 4 years. Smoking was linked to death.
No differenceNo disclosure in PubMedAbstract onlyNo disclosure in PubMed.
-
Post-diagnosis dietary factors, supplement use and colorectal cancer prognosis: A Global Cancer Update Programme (CUP Global) systematic literature review and meta-analysis.
40 studies, 30,242 people · People who already had bowel cancer, from 5 trials and 35 follow-up studies
TestedLooked at many foods and supplements after diagnosis, including coffee, and later death
Total, caffeinated and decaf coffee went with lower death from any cause in bowel cancer survivors. The review called the evidence limited-suggestive. No coffee-specific numbers in the abstract.
Points to benefitNo disclosure in PubMedAbstract onlyNo disclosure in PubMed; funded by WHO and cancer research charities.
-
Coffee consumption and all-cause and cause-specific mortality: a meta-analysis by potential modifiers.
40 studies, 3,852,651 people · Adults in 40 studies that followed people over time
TestedCompared people who drank different amounts of coffee with those who drank none
Risk of death from any cause was lowest at 3.5 cups a day (RR = 0.85, 95% CI 0.82-0.89). More cups did not lower it further. The link held for age, weight, smoking and decaf.
Points to benefitNo disclosure in PubMedAbstract onlyNo disclosure in PubMed.
-
Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes.
218 studies · Adults in many countries, from earlier pooled studies on coffee and health
TestedReview of 201 pooled analyses of observational studies and 17 pooled analyses of trials; compared high vs low coffee, any vs none, and one extra cup a day
Drinking three to four cups a day vs none went with a lower risk of death from any cause (relative risk 0.83, 95% confidence interval 0.79 to 0.88). Heart disease death was also lower (0.81, 0.72 to 0.90).
Points to benefitNo stake declaredAbstract onlySome authors report fees from drug companies, not coffee makers; no support for this work.
-
Coffee's Impact on Health and Well-Being.
Adults, from long-term follow-up studies around the world
TestedA look back at research on moderate coffee drinking, mostly 3 to 5 cups a day
The review says large follow-up studies show moderate coffee, typically three to five cups a day, is linked to lower overall death. Added sugar and cream may weaken the benefit.
No weight on the ladderCompany stakeAbstract onlyOne author has been a consultant for the National Coffee Association; the paper says it was not funded by it.
-
Opinion or narrative reviewMattioli, 2025Nutrition, metabolism, and cardiovascular diseases : NMCD · PubMed
Cardiovascular health and coffee: Insights from a recent cohort study.
General adults, especially people with heart disease
TestedCommentary on a cohort study of 1 to 3 cups of coffee a day
Commentary says 1-3 cups a day goes with lower death from any cause and heart disease, and calls for more research. It reports no new data.
No weight on the ladderNo disclosure in PubMedAbstract onlyNo disclosure in PubMed.
-
Coffee consumption and cardiometabolic health: a comprehensive review of the evidence.
Adults, from long-term follow-up studies, trials and gene-based studies
TestedA look back at studies on coffee and heart and metabolic health, 1 to 5 cups a day
Coffee goes with lower risk of death from any cause, with the biggest drop at moderate amounts (U-shaped). Gene-based studies do not clearly show a causal harm or benefit.
No weight on the ladderNo stake declaredAbstract onlyNo competing interests declared.
-
Opinion or narrative reviewBarrea et al., 2023Critical reviews in food science and nutrition · PubMed
Coffee consumption, health benefits and side effects: a narrative review and update for dietitians and nutritionists.
Adults, from mostly survey and follow-up studies
TestedA look back at coffee and health, with advice on amounts
Coffee seems linked to fewer cancers and lower risk of death from any cause. Up to 400 mg/day of caffeine (1-4 cups a day) looks safe. Trials are needed.
No weight on the ladderNo disclosure in PubMedAbstract onlyNo disclosure in PubMed.
-
Coffee for Cardioprotection and Longevity.
General adults
TestedReview of research on coffee and health, mostly observational
Large studies consistently link coffee to lower death from any cause. Drinking 3 to 4 cups regularly looked safe and went with the most benefit. Most data are observational, with very few randomized trials.
No weight on the ladderNo disclosure in PubMedAbstract onlyNo disclosure in PubMed.
38 papers we read and set aside as off topic
- Grosso et al., 2017. Coffee, Caffeine, and Health Outcomes: An Umbrella Review. (systematic review)
- Ding et al., 2014. Long-term coffee consumption and risk of cardiovascular disease: a systematic review and a dose-response meta-analysis of prospective cohort studies. (pooled review (meta-analysis))
- Chapelle et al., 2020. Recent advances in clinical practice: colorectal cancer chemoprevention in the average-risk population. (systematic review)
- Onyije et al., 2024. Risk factors for childhood brain tumours: A systematic review and meta-analysis of observational studies from 1976 to 2022. (pooled review (meta-analysis))
- Kennedy et al., 2016. Systematic review with meta-analysis: coffee consumption and the risk of cirrhosis. (pooled review (meta-analysis))
- Alfaro et al., 2018. Chronic coffee consumption and respiratory disease: A systematic review. (systematic review)
- Shaban et al., 2026. Coffee, caffeine, and cardiovascular health: navigating risks and benefits-an updated systematic review and meta-analysis. (pooled review (meta-analysis))
- Groll et al., 1986. Cimetidine prophylaxis for gastrointestinal bleeding in an intensive care unit. (randomized trial)
- Tresserra-Rimbau et al., 2013. Dietary intake and major food sources of polyphenols in a Spanish population at high cardiovascular risk: the PREDIMED study. (population study: one-time survey)
- Momas et al., 1994. Relative importance of risk factors in bladder carcinogenesis: some new results about Mediterranean habits. (population study: cases vs controls)
- Uto-Kondo et al., 2010. Coffee consumption enhances high-density lipoprotein-mediated cholesterol efflux in macrophages. (cell study)
- Lin et al., 1990. A controlled study of therapeutic endoscopy for peptic ulcer with non-bleeding visible vessel. (randomized trial)
- Rockey et al., 2017. Randomized pragmatic trial of nasogastric tube placement in patients with upper gastrointestinal tract bleeding. (randomized trial)
- Shaukat et al., 2017. BMI Is a Risk Factor for Colorectal Cancer Mortality. (population study: followed a group)
- Kondo et al., 2021. Association between coffee and green tea intake and pneumonia among the Japanese elderly: a case-control study. (population study: cases vs controls)
- Niv et al., 2002. Protective effect of faecal occult blood test screening for colorectal cancer: worse prognosis for screening refusers. (trial, not randomized)
- Schumacher et al., 2011. Thrombosis preventive potential of chicory coffee consumption: a clinical study. (trial, not randomized)
- Rubin et al., 2011. Live view video capsule endoscopy enables risk stratification of patients with acute upper GI bleeding in the emergency room: a pilot study. (randomized trial)
- Loftfield et al., 2015. Associations of Coffee Drinking with Systemic Immune and Inflammatory Markers. (population study: one-time survey)
- Muftic et al., 2013. Osteoporosis--comparative study between quantitative ultrasound of calcaneus and DXA. (population study: one-time survey)
- Bertoia et al., 2013. Long-term alcohol and caffeine intake and risk of sudden cardiac death in women. (population study: followed a group)
- Richardson et al., 2009. Randomized control trial investigating the influence of coffee on heart rate variability in patients with ST-segment elevation myocardial infarction. (randomized trial)
- Yeung et al., 2023. Low Urinary Potassium Excretion Is Associated with Higher Risk of All-Cause Mortality in Patients with Type 2 Diabetes: Results of the Dutch Diabetes and Lifestyle Cohort Twente (DIALECT). (population study: followed a group)
- Kruse et al., 2016. Does caffeine reduce postoperative bowel paralysis after elective laparoscopic colectomy? (CaCo trial): study protocol for a randomized controlled trial. (trial plan, no results yet)
- Czepa et al., 2014. [Food consumption and dietary behaviour in haemophilia patients before and after a nutrition consultation]. (randomized trial)
- Poikolainen et al., 2008. Magnesium treatment in alcoholics: a randomized clinical trial. (randomized trial)
- Zhang et al., 2026. Coffee and Tea Intake, Dementia Risk, and Cognitive Function. (population study: followed a group)
- Kunutsor et al., 2025. Coffee consumption, cancer, and healthy aging: epidemiological evidence and underlying mechanisms. (opinion or narrative review)
- Saad et al., 2025. Theobromine is associated with slower epigenetic ageing. (blood level study)
- Hak et al., 2008. Lifestyle and gout. (opinion or narrative review)
- Zhang et al., 2024. Associations between different coffee types, neurodegenerative diseases, and related mortality: findings from a large prospective cohort study. (population study: followed a group)
- Bielecka et al., 2020. The Influence of Nutritional and Lifestyle Factors on Glioma Incidence. (opinion or narrative review)
- Inoue-Choi et al., 2022. Tea Consumption and All-Cause and Cause-Specific Mortality in the UK Biobank : A Prospective Cohort Study. (population study: followed a group)
- Tabara et al., 2019. Lifestyle habits associated with nocturnal urination frequency: The Nagahama study. (population study: one-time survey)
- Nivukoski et al., 2020. Combined effects of lifestyle risk factors on fatty liver index. (population study: one-time survey)
- Kaplan et al., 2019. Smoke-free Turkey: Evaluation of outdoor areas of public places. (population study: one-time survey)
- Singh et al., 2014. Chemopreventive strategies in hepatocellular carcinoma. (opinion or narrative review)
- Shirai et al., 2020. Green tea and coffee intake and risk of cognitive decline in older adults: the National Institute for Longevity Sciences, Longitudinal Study of Aging. (population study: followed a group)
Words on this page
- Hazard ratio (HR)
- Compares how often something happened in two groups over time. 1 means no difference. 0.80 means 20% lower risk; 1.20 means 20% higher.
- Relative risk (RR)
- Compares the risk in two groups. 1 means no difference. 0.85 means 15% lower risk.
- 95% CI (confidence interval)
- The range where the true result likely lies. If the range includes no effect (1 for ratios, 0 for differences), the result is not clear.
- Pooled review (meta-analysis)
- A study that combines the results of many earlier studies to get one overall answer.
- Cohort study
- A study that follows a large group of people over years. It shows links, not causes.
How the short answer is set
Fixed rules turn the studies into the short answer, so the same studies always give the same answer. The rules start from GRADE, a standard system for rating evidence, in plain language.
| Confidence | What it means |
|---|---|
| High | We are confident the answer is right. |
| Moderate | The answer is probably right, but it could change. |
| Low (this page) | The answer may well change. |
| Very low | Any answer is very uncertain. |
| Not tested in people | Only cell or animal studies exist. |
Why low here
- All evidence in people comes from population studies, which show links, not causes.
- Results point the same way in large US, UK, European and Asian studies.
- The lower risk is modest and does not keep growing past a few cups.
- No trial among these papers measured death from any cause.
Hard limit used. With no human trial of death from any cause, the answer cannot say coffee makes people live longer.
How this page was made
Label
- Search
- PubMed, Nov 14, 2026. Searches written by Claude Opus 5.5; papers ranked by PubMed
- First read of all 53 papers
- Claude Sonnet
- Independent second read
- Claude Haiku
- Disagreements settled by
- Claude Opus 5.5
- Page text written by
- Claude Opus 5.5
- Charts
- Drawn by code from the study records
- Source check by a person
- Not yet. Required before search engines may list this page.
- Expert review
- No
- Papers read in full
- 0 of 53; all from PubMed abstracts
- Method
- v0.2, created Nov 14, 2026
Search. We ran four PubMed searches for coffee and death from any cause: pooled and systematic reviews since 2012 (10 of 56), trials in people (21 of 21), population studies (22 of 333). The first number is how many papers we read; the second is how many PubMed found. We took the ones PubMed ranks as most relevant. See the full PubMed search. Of the 53 papers read, 15 were on topic.
Limits. Digestery reads at most about 60 papers per question, so large fields are sampled, not covered in full. Abstracts can leave out key details, such as side effects and funding.
Review queue
These items wait for a person. Search engines are asked not to list the page until the first one is done.
- OpenThe short answer is new. It waits for a source check by a person.
- OpenGrosso 2017: abstract says 'cardiovascular disease and mortality'; unclear if this means death from any cause. Marked off-topic; check full text.
- OpenNo COI statement in PubMed for Loftfield 2015, Kim 2019, Chan 2024, Martin 1988, Barrea 2023, Liu 2022, Mattioli 2025 and O'Keefe 2018; check full texts for disclosures.
- OpenPoole 2017 and Ma 2023 authors have drug, supplement or walnut ties, not coffee; marked no stake. Confirm this reading.
- OpenEmadi 2025 author consulted for the National Coffee Association; paper says it was not funded or reviewed by it. Confirm the company label.
- OpenKim 2019 and Poole 2017 may pool overlapping cohorts, possibly including Loftfield 2015; avoid counting them as independent evidence.
- OpenZhou 2025 and Yan 2025 both use US NHANES data (1999-2018 and 2001-2018); people likely overlap, so not independent evidence.
- OpenMattioli 2025 is a commentary on a Zhou et al. cohort stressing people with heart disease; check whether it is Zhou 2025 or a missing paper.
- OpenYan 2025 reports modelled life expectancy, not hazard ratios; its 2·02-year gain depends on the paper's methods.
- OpenIten 2024: death from any cause is a secondary outcome with no number in the abstract; check the full text.
- OpenMartin 1988 measured total caffeine from coffee, tea and medicines, not coffee alone; PubMed labels it a trial but it is a cohort.
Page history
| Date | Change | Reason |
|---|---|---|
| Nov 14, 2026 | Page created from 53 papers. | First version of the page. |
Flag an error
A flag must point to a source: the passage in the paper, or a link to a study this page missed. A flag starts a new check. Votes never change a page.