How Much Coffee Is Too Much? A 498,134-Person Study Says Probably More Than You Think
Share
Short answer: for most healthy adults, there is no good evidence that normal coffee drinking shortens your life. In the largest study ever run on the question, the people drinking eight or more cups a day had a lower risk of dying over the next ten years than the people who drank none.
That is not a line we invented to sell coffee. It comes from a study of 498,134 adults published in JAMA Internal Medicine, and it is probably the single most useful piece of research a coffee drinker can know about.
Here is what it found, what it does not prove, and what to actually do with it.
The study: 498,134 people, ten years, 14,225 deaths
Researchers at the National Cancer Institute used the UK Biobank, a database of roughly half a million British adults who signed up between 2006 and 2010 and agreed to be tracked. The team, led by Erikka Loftfield, pulled everyone with complete data on coffee intake and smoking status. That left 498,134 people. Mean age 57. Fifty four percent women. Seventy eight percent of them drank coffee.
Then they waited. Over roughly ten years of follow up, 14,225 of those people died. The question was simple: did the coffee drinkers die at a different rate than the non drinkers?
They did. They died less.
The numbers, cup by cup
Below are the hazard ratios for all cause death, using non coffee drinkers as the reference group. A hazard ratio under 1.00 means a lower risk than the non drinkers. The numbers in brackets are the 95 percent confidence interval, which is the range the true value most likely sits in.
| Coffee per day | Hazard ratio for death | 95% CI |
|---|---|---|
| None | 1.00 (reference) | n/a |
| Less than 1 cup | 0.94 | 0.88 to 1.01 |
| 1 cup | 0.92 | 0.87 to 0.97 |
| 2 to 3 cups | 0.88 | 0.84 to 0.93 |
| 4 to 5 cups | 0.88 | 0.83 to 0.93 |
| 6 to 7 cups | 0.84 | 0.77 to 0.92 |
| 8 or more cups | 0.86 | 0.77 to 0.95 |
Read the bottom row again. Eight or more cups a day, and the risk of dying was still 14 percent lower than drinking none.
One honest note on the top row: the "less than 1 cup" group has a confidence interval that crosses 1.00, which means that result is not statistically significant on its own. Every other row is.
The part nobody expected: the caffeine gene did not matter
This is the reason the study was run in the first place, and it is the finding that makes it different from every coffee study before it.
People metabolize caffeine at wildly different rates, and a lot of that is genetic. Variants in four genes, AHR, CYP1A2, CYP2A6 and POR, change how fast your body clears caffeine. The long standing worry was that slow metabolizers were quietly paying a price for their coffee habit. Caffeine sits in their system longer. Surely that shows up somewhere.
Because UK Biobank had genetic data on these participants, the researchers could build a caffeine metabolism score and split the group by it. Fast metabolizers. Slow metabolizers. Everyone in between.
The protective association held across all of them. For people drinking six or more cups a day, hazard ratios across the genotype strata ran from 0.70 to 0.92, and there was no statistical evidence that genotype changed anything (P = 0.17 for heterogeneity).
If you have been told you are a slow caffeine metabolizer and should therefore be nervous about coffee, this is the largest dataset that has ever looked at that question, and it did not find the problem.
Decaf and instant showed the same pattern
The researchers broke the results out by type of coffee: ground, instant, and decaffeinated. All three showed similar inverse associations with mortality.
Decaf is the interesting one. If caffeine were doing the work, decaf should not show the same pattern. It did. The authors' own conclusion says the findings "suggest the importance of noncaffeine constituents in the coffee-mortality association."
Coffee is not a caffeine delivery system with some flavor attached. A roasted coffee bean contains hundreds of compounds, including chlorogenic acids, diterpenes, trigonelline and melanoidins formed during roasting. Something in that mix, or several somethings, is likely doing more than the stimulant is.
Does it replicate?
Yes, and at enormous scale.
A 2019 meta analysis in the European Journal of Epidemiology pooled 40 studies covering 3,852,651 people and 450,256 deaths. It found the same inverse relationship, with the lowest risk landing around 3.5 cups per day for all cause mortality (RR 0.85, 95% CI 0.82 to 0.89). The association held regardless of age, overweight status, alcohol intake, smoking status, and the caffeine content of the coffee.
Roughly four million people, two independent research groups, same direction.
What this does not prove
We are going to be straight about this, because a study this good deserves to be described accurately.
- It is observational, not a randomized trial. Nobody assigned people to drink eight cups a day. Coffee drinking is associated with lower mortality. That is not the same as coffee causing lower mortality.
- Reverse causation is a real concern. People who feel unwell often cut back on coffee. That alone could make non drinkers look sicker. The researchers adjusted for known health conditions, but adjustment is never perfect.
- UK Biobank volunteers are healthier than the general population. People who sign up for a decade of health tracking are not a random slice of humanity.
- The absolute differences are small. A 14 percent lower relative hazard over ten years is meaningful at population scale and modest for any one person.
- Several researchers pushed back. JAMA Internal Medicine published four separate letters challenging the interpretation, and the authors replied to them. That argument is part of the record and worth knowing about.
What the study genuinely supports is the authors' own careful phrasing: this provides "further reassurance that coffee drinking can be a part of a healthy diet." Not a cure. Not a supplement. Reassurance.
So how much should you actually drink?
The pooled data puts the bottom of the curve somewhere around two to four cups a day, and drinking more than that did not make things worse in either dataset.
Three things the mortality data cannot tell you, which still matter:
- Sleep. This study looked at death over ten years, not at how you slept last night. Caffeine has a half life of roughly five to six hours in most adults. Eight cups spread across a morning is a different thing than eight cups ending at 8pm.
- What you put in it. None of this is a case for a 600 calorie sugar drink with coffee in it.
- You. Pregnancy, arrhythmias, reflux, anxiety and certain medications all change the calculation. Population averages are not medical advice for one person.
Why this matters if you train
Most of the caffeine research aimed at athletes is about the next 60 minutes. Does it make the set faster, the effort feel easier, the time trial quicker. Useful questions, and the answer is generally yes.
This study is asking a different one. Over ten years, is the habit costing you anything?
The largest dataset we have says no, and says it across every caffeine metabolism genotype they could measure. For anyone who drinks coffee every training day and has quietly wondered whether they should cut back, that is the number to know.
Drink good coffee. PR Blend is the one we built for training days, and the rest of the Caffeine and Kilos coffee lineup is roasted to order and shipped fresh.
Frequently asked questions
How much coffee is too much?
In the UK Biobank data, even eight or more cups a day was associated with lower all cause mortality than drinking none. The pooled meta analysis puts the lowest risk around two to four cups a day. The practical ceiling for most people is set by sleep and tolerance, not by mortality risk.
Is coffee bad for you if you are a slow caffeine metabolizer?
The study specifically tested this using a genetic score built from variants in AHR, CYP1A2, CYP2A6 and POR. The lower mortality held across fast and slow metabolizers alike, with no statistical evidence of a difference (P = 0.17).
Does decaf count?
In this study, yes. Decaffeinated, instant and ground coffee all showed similar associations, which is a strong hint that the benefit is not coming from the caffeine.
Does coffee cause you to live longer?
No study has shown that, and this one cannot. It is observational. It shows an association, consistently, across millions of people, which is meaningful but is not proof of cause.
Is instant coffee as good as fresh ground?
For the mortality association in this study, instant showed a similar pattern to ground. For taste, freshness and how much you actually enjoy drinking it, that is a different question with a different answer.
Sources
- 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
- Kim Y, Je Y, Giovannucci E. Coffee consumption and all-cause and cause-specific mortality: a meta-analysis by potential modifiers. European Journal of Epidemiology. 2019;34(8):731-752. doi:10.1007/s10654-019-00524-3
- Correspondence on the Loftfield findings and the authors' reply, JAMA Internal Medicine. 2018;178(12):1723-1727.
This article is about population level research. It is not medical advice, and it is not a substitute for talking to your doctor about your own situation.