A coffee bean is a seed. It comes from the fruit of the coffee plant, called a coffee cherry, and each cherry usually holds two seeds, though 5 to 10% hold only one. Arabica, the source of about 60% of the world’s coffee products, is itself a hybrid of two wild species, Coffea canephora (robusta) and Coffea eugenioides, which crossed 350,000 to 610,000 years ago. Robusta beans carry about 40 to 50% more caffeine than arabica.
A brewed cup (237 g) is 99% water and has 2 kcal and about 95 mg of caffeine. How you brew it matters more for your cholesterol than most people expect. Cafestol, a compound in coffee, has been called the most potent cholesterol-raising compound known in the human diet. It is found in unfiltered brews such as boiled, Turkish and cafetière coffee, and in trials filtered coffee barely changed cholesterol.
Most of coffee’s good reputation comes from large cohort studies. They link 3 to 4 cups a day to a lower risk of dying, type 2 diabetes, liver disease and Parkinson’s disease, but they cannot prove cause, and genetic studies mostly do not back the heart and diabetes links. Trials are fewer and shorter. One found fewer relapses of atrial fibrillation in people who kept drinking coffee.
Health benefits of coffee
The best-supported finding is about brewing: in a meta-analysis of 14 trials, unfiltered or boiled coffee raised LDL cholesterol while filtered coffee barely did. In trials, regular coffee did not make atrial fibrillation return more often (it returned less often in one), and over weeks it did not raise blood pressure, although a dose of caffeine does for a few hours. The bigger claims, a lower risk of death, diabetes, liver cancer and cirrhosis, rest on observational studies, and genetic studies mostly do not support them. Here is what 30 studies found, strongest evidence first, from cholesterol and heart rhythm to blood pressure, pregnancy, liver disease and survival, including the findings that cut the other way.
Drinking coffee for weeks raised total and LDL cholesterol when it was unfiltered or boiled, but filtered coffee barely changed it
Meta-analysis of 14 randomized trials of coffee drinking. Cholesterol rose in step with the amount of coffee, and the rises were larger in people with high blood fats and in trials of caffeinated or boiled coffee. The abstract gives no pooled numbers.
In people with atrial fibrillation, drinking about a cup of caffeinated coffee a day for 6 months gave a 39% lower hazard of the rhythm problem returning than giving up coffee
Randomized trial in 200 coffee drinkers (average age 69) who were scheduled for electrical cardioversion for persistent atrial fibrillation or flutter. The rhythm problem returned in 47% of the coffee group and 64% of the abstinence group (hazard ratio 0.61). People knew which group they were in, and side effects did not differ significantly.
Drinking coffee for about 45 days raised LDL cholesterol by 5.4 mg/dL and total cholesterol by 8.1 mg/dL on average
Meta-analysis of 12 randomized trials in Western countries (1,017 people). Triglycerides rose 12.6 mg/dL. The rise in total cholesterol was bigger with unfiltered and caffeinated coffee, in people with high blood fats, and with more coffee.
Cutting caffeine by about 182 mg a day in pregnancy made no significant difference to birth weight or length of pregnancy
Double-blind randomized trial in 1,207 pregnant Danish women who drank at least 3 cups of coffee a day, given caffeinated or decaffeinated instant coffee. Babies in the decaf group weighed 16 g more and pregnancies ran 1.31 days shorter, both within chance. Only a moderate reduction in caffeine was tested.
Show 26 more studies
In people with high blood pressure, 200 to 300 mg of caffeine raised systolic pressure by 8.1 mm Hg for at least 3 hours but 2 weeks of coffee did not raise it
Meta-analysis of 5 trials of caffeine doses, 3 trials of 2 weeks of coffee against decaf or no coffee, and 7 cohort studies. Diastolic pressure rose 5.7 mm Hg after the caffeine doses. Habitual coffee drinking was not linked to more heart disease in hypertension.
In the CRAVE trial, caffeinated coffee did not significantly raise premature atrial contractions (58 vs 53 a day) over 2 weeks, but premature ventricular contractions rose from 102 to 154 a day
Randomized trial in 100 adults (average age 39) with heart monitors, told by daily text to drink caffeinated coffee or avoid caffeine across 14 days. On coffee days people took 1,058 more steps and slept 36 fewer minutes, and blood sugar did not differ. It was small and short and measured heartbeats, not heart disease.
Five to six cups a day of cafetière coffee for 24 weeks raised LDL cholesterol by 0.26 mmol/L compared with filtered coffee
Randomized trial in 46 healthy adults (22 on cafetière, 24 on filtered coffee) drinking 0.9 liter of strong coffee a day. The liver enzyme ALT stayed 45% higher with cafetière coffee and passed the normal limit in 8 of 22 people. All changes reversed after the coffee was stopped.
Across 15 trials and cohort studies, regular coffee drinking showed no significant effect on blood pressure or on the risk of hypertension
Systematic review of randomized trials and cohort studies lasting at least 1 week. Pooled systolic pressure changed by -0.55 mm Hg and diastolic by -0.45 mm Hg, neither significant, and the pooled risk ratio for hypertension was 1.03. The authors rated the evidence low quality and made no recommendation for or against coffee.
Five cups of coffee a day for 8 weeks did not change blood sugar control or insulin sensitivity in overweight adults
Randomized trial in 45 healthy overweight regular coffee drinkers, given caffeinated coffee, decaf or water. Caffeinated coffee raised adiponectin by 1.4 µg/mL and interleukin-6 by 60%, and decaf lowered fetuin-A by 20%. The trial was small and short.
An umbrella review found the lowest risk of death at 3 to 4 cups a day (relative risk 0.83) across mostly observational studies
Umbrella review of 201 meta-analyses of observational studies and 17 of interventional studies. At 3 to 4 cups a day the relative risk was 0.85 for cardiovascular disease, and high versus low intake went with an 18% lower risk of cancer. Harm showed up mainly in pregnancy, and the authors call for trials to learn whether the links are causal. No industry support was declared.
In 521,330 Europeans followed for 16.4 years, the highest coffee drinkers had a lower risk of dying (hazard ratio 0.88 in men, 0.93 in women)
Cohort from 10 countries with 41,693 deaths, compared with non-drinkers. Digestive disease deaths were lower too (0.41 in men, 0.60 in women), while ovarian cancer deaths were higher in women (1.31). Coffee habits were asked only once, and results held after excluding people who died within 8 years.
Among 185,855 people from five ethnic groups, 2 to 3 cups a day went with a lower risk of death (hazard ratio 0.82)
Multiethnic Cohort in Hawaii and Los Angeles, with 58,397 deaths over an average of 16.2 years, compared with no coffee. One cup a day gave 0.88 and 4 or more cups 0.82. Trends were similar for caffeinated and decaf coffee and in never-smokers, but the link was not significant in Native Hawaiians.
In 498,134 UK adults, 2 to 3 cups a day went with a lower risk of death (hazard ratio 0.88) whether people broke down caffeine slowly or quickly
UK Biobank cohort with 14,225 deaths over 10 years, compared with non-drinkers. Ground, instant and decaf coffee showed similar links, and 8 or more cups a day still gave 0.86. A genetic score for caffeine metabolism did not change the pattern, which points to parts of coffee other than caffeine.
Genetic (Mendelian randomization) studies did not support cause for coffee’s link to lower death and heart disease risk except for gallstones
Review that set Mendelian randomization studies of coffee against observational ones. Hazard ratios of 0.85 to 0.90 for death and about 0.70 for type 2 diabetes in coffee drinkers were not backed by the genetic studies. They did support a lower gallstone risk (0.81) and hinted at a higher colorectal cancer risk.
Unsweetened coffee went with a lower risk of death (hazard ratios 0.71 to 0.84) in 171,616 UK adults, and sweetened coffee only up to a point
UK Biobank cohort with 3,177 deaths over a median of 7 years. Sugar-sweetened coffee gave 0.69 at 1.5 to 2.5 drinks a day but 1.05, no link, above 4.5 drinks. Results for artificially sweetened coffee were less consistent, and the curves were U-shaped.
Across 36 cohort studies, 3.5 cups a day went with a relative risk of cardiovascular disease of 0.85 while 5 cups showed no clear drop (0.95)
Dose-response meta-analysis of 1,279,804 people and 36,352 cardiovascular disease cases. Risk was lowest at 3 to 5 cups a day, and heavy drinking was not linked to higher risk. Cohort studies cannot show cause.
In 28 cohort studies, each extra cup of coffee a day went with a relative risk of type 2 diabetes of 0.91 for caffeinated coffee and 0.94 for decaf
Dose-response meta-analysis of 1,109,272 people and 45,335 diabetes cases. Compared with no or rare coffee, the relative risk fell from 0.92 at 1 cup to 0.67 at 6 cups a day. Both caffeinated and decaf coffee showed the pattern, and it does not prove that coffee prevents diabetes.
Each extra 2 cups of coffee a day went with a 35% lower risk of liver cancer (relative risk 0.65)
Meta-analysis of 18 cohorts (2,272,642 people, 2,905 cases) and 8 case-control studies. Cohorts gave 0.71, and per 2 cups the risk was 0.73 for caffeinated coffee and 0.86 for decaf. The authors rated the evidence very low quality under GRADE, citing no trials and possible publication bias.
Each extra 2 cups of coffee a day went with about half the risk of cirrhosis (relative risk 0.56)
Meta-analysis of 5 cohort and 4 case-control studies (1,990 cases, 432,133 people). Cohorts gave 0.58 and case-control studies 0.52, and alcoholic cirrhosis 0.62. Results varied a lot between studies (I² 83.3%), so the size of the effect is uncertain.
Across 10 prospective studies, each extra cup of coffee a day went with a 2% lower risk of atrial fibrillation
Dose-response meta-analysis of 723,825 people with 30,169 atrial fibrillation events. Risk at every amount from 1 to 7 cups was not significantly different from none (0.87 at 7 cups), so the trend is toward lower risk, not higher.
In 6 cohorts, 1 to 3 cups a day went with a slightly higher risk of hypertension (relative risk 1.09) while more than 5 cups did not
Meta-analysis of 172,567 people and 37,135 new cases of hypertension, followed for 6.4 to 33 years, compared with under 1 cup a day. The curve was J-shaped: risk rose up to 3 cups and was 0.99 at 6 cups.
Each extra 100 mg of caffeine a day in pregnancy went with a 14% higher risk of miscarriage and a 7% higher risk of low birth weight
Meta-analysis of 60 publications from 53 cohort and case-control studies. Stillbirth risk was 19% higher and small-for-gestational-age 10% higher, while preterm delivery was not significantly different. Studies varied a lot, no safe threshold was found, and the authors say the effects are modest and could come from bias.
Caffeine of 350 to 699 mg a day in pregnancy went with a 40% higher risk of pregnancy loss
Meta-analysis of 14 prospective studies (130,456 women, 3,429 cases). Risk was 1.02 at 50 to 149 mg and 1.16 at 150 to 349 mg, neither significant, and 1.72 at 700 mg or more. Each 100 mg was linked to 7% higher risk. Publication bias and unmeasured smoking may have played a part.
Women with the highest blood caffeine levels in early pregnancy had babies who weighed 84 g less at birth than women with the lowest levels
US cohort of 2,055 nonsmoking women at low risk. Reported intake of about 50 mg a day, half a cup of coffee, went with babies 66 g lighter. Length and head size were smaller too, and results did not differ by caffeine-metabolism genes. Observational, so other factors may play a part.
People who drank the most coffee had a lower risk of depression (relative risk 0.76)
Meta-analysis of 12 observational studies (346,913 people, 8,146 cases of depression). The curve was J-shaped, with the biggest protection near 400 mL a day. Observational, so it does not show that coffee prevents depression.
About 3 cups of coffee a day went with a lower risk of Parkinson’s disease (smoking-adjusted relative risk 0.72)
Dose-response meta-analysis of 13 articles and 901,764 people. Protection leveled off at about 3 cups a day, was stronger in men than in women, and was weaker in the US than in Europe or Asia. Further studies are needed.
In a genetic study, a higher genetic tendency to drink coffee went with 2.79 times the odds of esophageal cancer but not with cancer overall
Mendelian randomization using UK Biobank data on 59,647 cancer cases and 22 site-specific cancers. Overall cancer was not significant (odds ratio 1.05). Multiple myeloma odds were higher (2.25) and ovarian cancer odds lower (0.63). The authors say the mechanisms need study.
Genetically predicted coffee intake was not linked to any of 15 cardiovascular conditions in a Mendelian randomization study
Used UK Biobank (up to 35,979 cases) and FinnGen (up to 17,325 cases). Odds per 50% genetic rise ranged from 0.97 for intracerebral hemorrhage to 1.26 for deep vein thrombosis in UK Biobank, and one adjusted FinnGen analysis hinted at higher hemorrhage risk (1.91). The authors suggest earlier observational links may have been confounded.
In mice, cafestol from unfiltered coffee raised blood cholesterol by 61% on a high dose by shutting down bile acid production
Mice fed cafestol for 8 weeks at amounts equal to 8 or 40 cups of unfiltered coffee a day in people, scaled to energy intake. Cholesterol rose 33% on the low dose. Cafestol cut activity of the main bile acid enzyme by 57%, which may explain the effect in people.
In cells and mice, cafestol switched on two receptors, FXR and PXR, that control cholesterol and bile acids
Cafestol lowered bile acid genes such as CYP7A1 in normal mice but not in mice lacking FXR. The authors describe cafestol as the most potent cholesterol-raising compound known in the human diet. A lab study of mechanism, not of human health.
Who should be careful with coffee
Most adults can drink coffee freely. Take a little care if one of these applies:
- Pregnancy. EFSA says caffeine up to 200 mg a day from all sources raises no safety concern for the fetus, and ACOG gives the same figure. That is a little over 2 cups of brewed coffee at about 95 mg a cup. The evidence above is mixed, so treat it as a sensible limit and ask your midwife or doctor.
- High LDL cholesterol. Choose filtered coffee. French press, boiled and Turkish coffee carry cafestol, which raised LDL by 0.26 mmol/L in a 24-week trial of 5 to 6 cups a day.
- Heart rhythm problems, high blood pressure or poor sleep. Caffeine lifts blood pressure for a few hours, and EFSA notes that even 100 mg close to bedtime can disturb sleep. Up to 400 mg a day, about 4 cups, is fine for healthy adults according to EFSA and FDA. If coffee brings palpitations, cut back and ask your doctor.
How to choose, store and eat coffee
- Filtered (drip, pour-over): filtered coffee showed very little cholesterol rise in a meta-analysis of 14 trials.
- Unfiltered (French press, boiled, Turkish): these keep cafestol, and 5 to 6 cups a day raised LDL by 0.26 mmol/L over 24 weeks.
- Decaf: decaf showed links to lower risk of death similar to regular coffee in the Multiethnic Cohort and UK Biobank, so it is a fair choice if caffeine bothers you.
Keep roasted beans or grounds in an opaque, airtight container in a cool, dark place, where opened beans stay fresh for 1 to 3 weeks and grounds for 1 to 2 weeks. In 2016 the IARC cancer agency rated drinking coffee as not classifiable as a cause of cancer (Group 3), but found that very hot drinks, above 65 °C, probably cause esophageal cancer, so let a fresh cup cool a little. Roasting forms acrylamide, which EFSA says may raise cancer risk across the whole diet, with limited and inconclusive human evidence; coffee is among the main sources, though fully roasted coffee holds only a small fraction of the acrylamide that forms early in roasting.
Sources
- EFSA Panel on Dietetic Products, Nutrition and Allergies. Scientific opinion on the safety of caffeine. EFSA Journal. 2015. With the explainer EFSA explains risk assessment: caffeine.
- ACOG Committee Opinion No. 462: Moderate caffeine consumption during pregnancy. Obstet Gynecol. 2010.
- Spilling the beans: how much caffeine is too much? U.S. Food and Drug Administration.
- Marcus GM, Hu FB, van Dam RM, et al. Caffeine and cardiovascular disease: a scientific statement from the American Heart Association. Circulation. 2026.
- IARC Monographs evaluate drinking coffee, maté, and very hot beverages (press release no. 244). International Agency for Research on Cancer. 2016.
- Loomis D, Guyton KZ, Grosse Y, et al. Carcinogenicity of drinking coffee, mate, and very hot beverages. Lancet Oncol. 2016.
- Acrylamide in food is a public health concern. European Food Safety Authority. 2015.
- Lantz I, Ternité R, Wilkens J, et al. Studies on acrylamide levels in roasting, storage and brewing of coffee. Mol Nutr Food Res. 2006.
- Storage and shelf life. National Coffee Association.
History of coffee
Coffee began as a wild plant in Ethiopia. A 2024 genome study found that arabica is a hybrid of two wild species and that the founding cross happened 350,000 to 610,000 years ago, long before anyone brewed it. Since then it has passed through several narrow population bottlenecks, which is why arabica varies so little genetically.
The famous story of a herder named Kaldi who discovered coffee does not appear in writing before 1671, and Wikipedia calls it likely apocryphal. There is no confirmed evidence of coffee as a drink before the 15th century. Sufi Muslims in Yemen used it in the late 15th century to stay focused during night prayers.
Coffee was not always welcome. A theological court in Mecca banned it in 1511 because of its stimulating effect, and Suleiman I overturned the ban in 1524. The chronicler İbrahim Peçevi records that two Arab merchants opened the first coffeehouse in Istanbul in 1554.
Coffee reached Europe through Italy and the Mediterranean in the mid-16th century. The first mainland European coffeehouse opened in Venice in 1645, and the first in England opened in Oxford in 1650.
Then plants began to travel. In 1616 the Dutch trader Pieter van den Broecke took coffee bushes from Mocha in Yemen to the botanical garden in Amsterdam. By 1719 the Dutch East India Company could supply Europe with Java coffee, and in 1720 Gabriel de Clieu brought seedlings to Martinique. Around 1727 Francisco de Melo Palheta smuggled seeds from French Guiana into Brazil, which became the world’s largest coffee producer by 1852 and has held that place since.
Sources
- Salojärvi J, Rambani A, Yu Z, et al. The genome and population genomics of allopolyploid Coffea arabica reveal the diversification history of modern coffee cultivars. Nat Genet. 2024.
- History of coffee. Wikipedia.
- Coffee. Wikipedia.
Frequently asked questions about coffee
Is coffee good for you?
For most adults, moderate coffee looks safe and may help, but the best evidence is for what it does not do. An umbrella review found the lowest risk of death at 3 to 4 cups a day (relative risk 0.83), yet that and most other benefits come from observational studies, and genetic studies mostly do not support cause. A cup (237 g) adds only 2 kcal.
How much coffee is too much?
EFSA and FDA treat up to 400 mg of caffeine a day as fine for healthy adults, which is about 4 cups at 95 mg a cup. In pregnancy the figure is 200 mg, a little over 2 cups. Large cohorts saw no extra risk of death even at 8 or more cups a day, but that is not a safety limit, and EFSA notes that 100 mg close to bedtime can disturb sleep in some adults.
Is coffee bad for your heart or blood pressure?
Not in the studies here, though caffeine has short-term effects. A dose of 200 to 300 mg of caffeine raised systolic pressure by 8.1 mm Hg for at least 3 hours in people with high blood pressure, but 2 weeks of coffee did not, and a pooled review found no significant effect on blood pressure. In the CRAVE trial coffee did not significantly raise premature atrial contractions but raised premature ventricular contractions from 102 to 154 a day, and in the DECAF trial atrial fibrillation returned in 47% of coffee drinkers against 64% of people who gave it up.
Does coffee raise cholesterol?
It can, depending on how it is brewed. Unfiltered or boiled coffee raised total and LDL cholesterol in a meta-analysis of 14 trials, and cafetière coffee raised LDL by 0.26 mmol/L against filtered coffee in a 24-week trial. Across 12 trials, coffee in general raised LDL by 5.4 mg/dL over about 45 days, while filtered coffee barely changed it.
Is decaf as healthy as regular coffee?
Decaf showed similar links in the large cohorts. Trends for death were similar for decaf and caffeinated coffee in 185,855 people, and the relative risk of type 2 diabetes per cup was 0.94 for decaf against 0.91 for caffeinated. These are observational results, and no trial here compared long-term health outcomes of decaf and regular. Decaf is a fair choice if caffeine bothers your sleep.
Is coffee safe during pregnancy?
Up to 200 mg of caffeine a day is the limit EFSA and ACOG give, a little over 2 cups of coffee. The evidence is mixed. Pooled observational studies linked each extra 100 mg to a 14% higher risk of miscarriage, and the highest blood caffeine levels went with babies 84 g lighter, but the one large trial in 1,207 women found no difference in birth weight or length of pregnancy when caffeine was cut by about 182 mg a day. Ask your midwife or doctor about your own limit.
Does coffee cause cancer?
There is no clear evidence that it does. In 2016 IARC rated drinking coffee as not classifiable (Group 3), and an umbrella review found an 18% lower risk of cancer at high versus low intake, with a relative risk of liver cancer of 0.65 per 2 cups. Those are observational results, and one genetic study found 2.79 times the odds of esophageal cancer. IARC also found that drinks hotter than 65 °C probably cause esophageal cancer, so let coffee cool a little.
How many calories and how much caffeine are in a cup of coffee?
A cup (237 g) of brewed coffee has 2 kcal, 0.3 g of protein, no carbohydrate and 94.8 mg of caffeine, or 40 mg per 100 g. It also gives 14% of the daily value of vitamin B2 (0.18 mg) and 116 mg of potassium. Milk and sugar add calories on top of that.