Large observational studies and meta-analyses consistently show an approximately 17% lower mortality risk at 3-4 cups of coffee per day compared to no coffee. The evidence is however almost exclusively observational; Mendelian randomization does not convincingly support causality for mortality in general. Exceptions apply to pregnant women and people with an increased risk of fractures or lung cancer.
Large observational studies consistently show that people who drink coffee daily have a lower risk of premature death than people who drink no coffee at all. An umbrella review of 201 meta-analyses (PMID 29167102) found a relative risk of 0.83 at 3-4 cups per day compared with no coffee: roughly 17% lower odds of dying. A separate meta-analysis of nearly 3.9 million participants (PMID 31055709) confirms this and shows that the most favourable point lies at around 3.5 cups per day. Drinking more beyond that threshold provides no additional benefit.
What you add to your coffee appears to matter. In a large American cohort study of 46,222 adults (PMID 40368300), the protective effect was measurable for black coffee and coffee with little sugar and saturated fat. Coffee with large amounts of cream and sugar showed no demonstrable benefit. In other words, it is not just the coffee itself that counts, but also what goes into it.
For people with type 2 diabetes there is specifically encouraging news: a Harvard cohort study of more than 15,000 diabetes patients with a mean follow-up of 18.5 years (PMID 37076174) found a 26% lower risk of death for high versus low coffee consumption. People who began drinking more coffee after their diabetes diagnosis also had lower mortality. A broad review in the journal Geroscience (PMID 38963648) confirms that the cardiometabolic benefits and the mortality benefit are most pronounced at moderate consumption, following a U-shaped relationship.
It is crucial to understand that all of this research is observational. Randomised controlled trials, the gold standard for establishing causality, are largely absent (PMID 29167102, 34455881, 38963648). People who drink a lot of coffee may also lead healthier lives in other ways. Mendelian randomisation, a technique that can provide indications of causality, supports a causal relationship convincingly only for type 2 diabetes and chronic kidney disease, not for all-cause mortality (PMID 38963648).
There are also risks and exceptions to keep in mind. Coffee consumption is consistently associated with an increased risk of lung cancer, even after researchers corrected for smoking (PMID 39266809). In women there is additionally an association with a higher risk of bone fractures, which is relevant for women who already have an elevated fracture risk (PMID 29167102). During pregnancy, moderation or avoidance is advisable: high consumption was associated with a 31% higher chance of low birth weight and a 46% higher chance of pregnancy loss (PMID 29167102). For healthy non-pregnant adults, up to 400 mg of caffeine per day, roughly 1-4 cups of coffee, is considered safe, although interactions with certain medications may play a role (PMID 34455881).
All claims are based on observational research (cohorts and meta-analyses). The largest review covers 201 meta-analyses and millions of participants, which makes the associations robust. Causality has not, however, been demonstrated through randomised trials; Mendelian randomisation supports causality only for type 2 diabetes and chronic kidney disease.