In large cohorts, 3 to 4 cups a day are associated with about 17% fewer deaths and less type 2 diabetes. In trials, unfiltered coffee raises cholesterol, while filtered coffee raises it very little. During pregnancy, caffeine is linked to more miscarriages.
“Strong” = several concordant meta-analyses or systematic reviews. “Moderate” = one synthesis, or several controlled trials. “Limited” = early trials only. The rating is about the quality of the studies, not about how useful the supplement is.

Bex Walton · CC BY 2.0
3 to 4 cups a day, the quantity linked to the largest risk reduction in cohorts; up to 8 cups a day tested in a trial.
Pregnancy: each 100 mg of caffeine per day (about one cup) is associated with a 7% higher risk of miscarriage, and 700 mg or more with 72% higher; high consumption is also linked to lower birth weights. Cholesterol: in 12 trials (1,017 people), 45 days of coffee raised total cholesterol by 8.1 mg/dL, mostly with unfiltered or caffeinated coffee and in people who already have too much cholesterol. Boiled or plunger coffee: in three volunteers, purified cafestol and kahweol raised cholesterol by 66 mg/dL in 6 weeks, and infusions containing them also raised triglycerides and a liver enzyme (ALT). Homocysteine: one litre of unfiltered coffee per day for two weeks increased it by 10%. Hypertension: in hypertensive individuals, 200 to 300 mg of caffeine raise blood pressure by 8.1 mmHg (systolic) and 5.7 mmHg (diastolic) for at least 3 hours; after two weeks of coffee, however, no increase. Sleep: caffeine reduces sleep duration and efficiency, and perceived quality. Fractures: the umbrella review finds a link between coffee and fractures in women, not in men. Limits: the harmful links mostly disappear once smoking is properly accounted for, except during pregnancy; and the favourable links also come from observational studies. In trials, 8 weeks of coffee changed neither glucose tolerance nor insulin sensitivity, and coffee low in caffeine and diterpenes changed neither lipids nor vessel function. Memory: out of 11 cohorts (29,155 people), coffee was not associated with less cognitive decline or dementia in general. Cancer: in the meta-analysis of 21 cohorts, coffee was not linked to cancer mortality. The link with deaths is clearer in Europe and Asia than in the United States.
What raises cholesterol in boiled coffee are two of its fats: in three volunteers, purified cafestol and kahweol increased it by 66 mg/dL in 6 weeks. Filtered coffee, on the other hand, raises it very little in trials.
This updated meta-analysis, based on more than 3.8 million subjects, examined the relationship between coffee consumption and mortality. It found non-linear inverse associations for all-cause, cardiovascular, and cancer mortality. Moderate coffee consumption, typically between 2 and 4 cups per day, was associated with a reduction in mortality risks. This inverse association for all-cause mortality was consistent regardless of age, overweight, alcohol consumption, smoking status, and the caffeine content of the coffee.
Kim Y, Je Y, Giovannucci E. — European journal of epidemiology (2019) — View the source study
For information only: not a substitute for medical advice. Dietary supplements are not a replacement for a varied, balanced diet. Learn more