The 20-minute version
It's ten past three on a Tuesday. You've been looking at the same paragraph for a few minutes without reading it. Your eyes are dry, there's a heaviness behind them and a slight fog over everything, and the meeting at four is going to need you sharp. So you make a coffee, the third of the day, and by twenty past you're back. The paragraph makes sense again.
The day started at half past six with the alarm, and the first thing you thought about was coffee. One at seven, standing up, before you could talk to anyone. Another at half past nine at your desk. If someone asked, you'd say two or three a day, and you wouldn't be counting the tea at eleven or the cola with lunch.
At half past five there's another, for the forty-minute drive home. The evening is fine. At eleven you go to bed, and at half past eleven you're still awake, not worried about anything in particular, just awake, with a mind that won't quite settle. Sleep comes some time after midnight. At half past six the alarm goes, you're tired, and you think about coffee.
You've noticed the shape of this. You may even have said it out loud. But the morning cup is the one thing in the day that works without fail, and giving it up, or even giving up the afternoon one, feels like choosing to be tired on purpose.
I'm not going to tell you coffee is bad for you. For most people at ordinary amounts the research says the opposite, and I'll come to it. I'm not asking you to give it up either. My argument is that most of what the cup gives you is the removal of something the previous cup caused, that what all of the cups take from the night is real and mostly invisible to you, and that you can keep the coffee you like for the reason you like it. The practical part is simple. You count what you actually have, you set a time after which you have none, you bring the total down gently over a couple of weeks, and you watch what your mornings do.
Three cautions first.
If you're pregnant or breastfeeding, the NHS advises no more than 200 milligrams of caffeine a day, which is roughly two mugs of instant coffee. That's a medical limit and it comes before anything here.
If caffeine gives you a racing or irregular heartbeat, or you've been told to limit it because of a heart condition, blood pressure, anxiety, a stomach condition or a medicine you take, follow that advice. The people treating you know your case and I don't.
And if you're exhausted every day despite sleeping seven hours or more, or you snore loudly and stop breathing in your sleep, or the tiredness has come with low mood, weight change or other symptoms, the problem may not be caffeine at all. See a GP before you change anything, and don't let the coffee hide the question.
One thing before you read on. For a week, write down every drink or food with caffeine in it and the time you had it, tea, cola, energy drinks and chocolate included. Each morning, note roughly when you fell asleep and how you feel out of ten. Don't change anything yet. You're only finding out what's true.
Back to the level
The cup works. Twenty minutes after it you feel better, and any argument that starts by calling that imagination deserves to lose you.
It's pleasant too. It smells good, it comes with a small ritual of kettle and mug and a minute by the window, and it's the one drug you can take in a meeting. It marks the start of the day, the middle of it and the end of dinner. Nearly everyone you know drinks it and says they need it. And the alternative seems to be sitting foggy at three with a meeting at four.
So the habit makes sense from inside. What I want to look at is what the cup is actually doing, and there's an experiment I'd ask you to read slowly, because the finding is easy to slide past.
Peter Rogers and colleagues at the University of Bristol took three hundred and seventy-nine people. Some were regular caffeine drinkers at medium to high levels, and some drank little or none. Everyone went sixteen hours without caffeine, which is roughly a night plus a morning. Then, under double-blind conditions, they got either caffeine or a placebo, and rated their alertness and headache before and after.
The regular drinkers who got the placebo felt worse, less alert and more headachy. The low and non-drinkers got no rise in alertness from caffeine at all. And when the regular drinkers got their caffeine, their alertness came back up to about where the non-drinkers already were, having had nothing. The researchers' own summary is that with frequent use, no net benefit for alertness is gained. Going without lowers it, and the caffeine merely returns it to baseline.
Now think about half past six.
The fog you wake into isn't your natural morning. It's what a regular drinker feels when yesterday's last dose has worn off overnight. The person opposite you on the train who has never drunk coffee doesn't feel it, because they woke up at the level. When you have your cup at seven you climb up to join them, and you don't pass them. You've spent twenty minutes and a coffee getting to where they started. You've been counting the morning cup as a gain, when it repays a loss you took on yesterday.
So "I can't function without my morning coffee" is true, and it's true because of the coffee. That isn't a criticism of you. It's what a daily dose of anything the body adapts to will do. It also means a non-drinker gets the morning coffee's effect every day, free, on waking.
"It gives me energy" goes the same way. A black coffee contains almost nothing your body can burn. As you stay awake, a substance called adenosine builds up in the brain and produces the feeling of sleepiness, and sleep clears it. Caffeine sits in the places where adenosine would act and blocks the signal. The adenosine is still there. You just stop feeling it. A message has been muffled.
And the message keeps arriving. Hans-Peter Landolt and colleagues in Zurich gave nine healthy men about two mugs' worth of caffeine at ten past seven in the morning and recorded their sleep that night. Compared with placebo they slept less and less efficiently, with less of the deep slow-wave sleep, and caffeine was still measurable in their saliva at eleven at night. A morning dose changed a night sixteen hours away.
So the pick-up is a loan. The sleepiness you'd have felt at three is deferred, and the deferral is paid for with a lighter night, which makes a bigger shortfall the next morning, which wants a bigger cup. The honest version of "it gives me energy" is "it stops me feeling how tired I am". Once in a while, before a long drive, that's useful. As a daily arrangement it means you never find out how tired you really are, so you never fix it.
What about thinking? A second Bristol study had regular drinkers and non-drinkers go without caffeine overnight, then tested them through the day. Regular drinkers on placebo got sleepier, slower and worse at remembering as the afternoon went on, and caffeine repaired that. In the non-drinkers it reduced sleepiness and did little else for the mind, partly because it made them jittery. The one thing it improved in everyone was physical speed. The authors called the paper "Faster but not smarter", and concluded that with regular use the daily cups don't improve mental performance. They only restore what going without took away. If the ideas come after the three o'clock cup, it's because the fog lifted, and the fog was the shortfall.
Which leaves three o'clock itself. Everyone flags after lunch, surely?
They do. Most people have a natural dip in alertness in the early afternoon, coffee or none, and no cup fixes that, because it isn't a shortfall. But in that second study the regular drinkers kept off caffeine got steadily worse into the afternoon. For you, that's the half past nine cup wearing off at the same hour as the natural dip. The two arrive together and you get the fog on the paragraph. The three o'clock cup removes the withdrawal part and leaves the natural part, which is why it never feels as complete as the one at seven. Bring the morning cups down and the withdrawal part shrinks, because there's less to withdraw from. What's left is a smaller, gentler dip you can deal with away from the kettle.
What the night pays
"It doesn't affect my sleep. I drop off fine."
This is the belief that lets the afternoon and evening cups happen. The test behind it is falling asleep, which is only one measure of a night, and you're in no position to judge the rest, because you were asleep for it.
Ian Clark and Hans-Peter Landolt reviewed the trials and population studies in 2017. Caffeine typically made people slower to fall asleep, and it also shortened sleep, made it less efficient and made people rate it worse. In the brain recordings there was less deep sleep and more light sleep, waking and arousal. Dose mattered and timing mattered. Older adults may be more sensitive, and people differed a great deal, partly for genetic reasons. The picture is a night that's shallower, more broken and a little shorter. From inside you can't feel any of it. You only have the morning, and you already have an explanation for the morning, which is that you're a tired person.
Timing is what people underestimate most. Christopher Drake and colleagues gave people 400 milligrams, about four mugs of instant, either at bedtime, three hours before it or six hours before it. All three disturbed sleep compared with placebo, including the dose six hours out. The NHS insomnia page says the same thing, no tea or coffee for at least six hours before bed. For an eleven o'clock bedtime, your half past five cup is inside that window and the three o'clock one is barely outside it.
You drop off fine because you're tired enough to drop off through the caffeine. That tells you how tired you are.
Then there's the cup nobody counts. The espresso after a meal out, or more often the mug of tea at nine on the sofa, made to wind down. Everything about that ritual says rest, and the contents say the opposite, about seventy-five milligrams of something whose job is to block the signal for sleep, two hours before you'd like to feel it. I believe the ritual relaxes you, so keep all of it, the kettle, the mug, the sofa, the ten minutes, and change what's in the mug. Decaf tea is close enough to tea that most people stop noticing within a week. A restaurant espresso on a special night is a fair trade if you make it knowingly. The same espresso every Friday, with a foggy Saturday morning for a year, is another thing. In my observation, the evening cup is the change the night notices soonest, often within three or four nights.
You may have tried cutting down before.
Within a day there was a headache like a band round your skull. You were foggy, flat and short-tempered, and by day three you'd concluded that your body needed coffee and gone back to it, feeling better within the hour. Laura Juliano and Roland Griffiths reviewed sixty-six studies of caffeine withdrawal in 2004. Headache, fatigue, low energy, drowsiness, low mood, poor concentration, irritability and fogginess were all well supported. About half the people in the experiments got a headache, and about one in eight had symptoms bad enough to interfere with their day. It began twelve to twenty-four hours after the last dose, peaked at twenty to fifty-one hours, and lasted between two and nine days. Going without as little as one mug of instant a day was enough to bring it on, and expectation wasn't what drove it.
So you were right about the headache. Look at the other number, though. Two to nine days. Your body has adapted to coffee and needs a few days to adapt back, and then the shortfall is gone for good and you wake at the level. Day three of your last attempt was the peak of a short process that you stopped just before it ended.
The same review found that avoiding withdrawal plays a central role in why people keep drinking caffeine. I think that's the heart of the habit. The morning cup lands just as the overnight withdrawal starts to bite, and the afternoon cup lands as the morning's wears off. Without anyone deciding it, the day is arranged so the shortfall is never felt in full, and the night pays for the arrangement. Nobody says you have to stop all at once, either. The method further on is built so the shortfall never gets deep.
"But coffee's good for you." Quite possibly. In 2017 Robin Poole and colleagues at Southampton pulled together two hundred and one meta-analyses of observational studies, and coffee was more often linked with benefit than harm, with the largest reductions in risk, including death from any cause and heart disease, at three to four cups a day. The exceptions were pregnancy, where higher intake went with low birth weight, preterm birth and pregnancy loss, and a link with fracture in women. The authors were clear that these are associations and that trials are needed to show cause. The European Food Safety Authority put the amount that raises no safety concern for healthy adults at up to 400 milligrams a day, with no more than 200 in one go, and 200 a day in pregnancy. For children and teenagers it found too little evidence to set a figure.
None of that research asked when people drank, or why, or how they slept. Three coffees before ten, and three at seven, three and half past five, give the same count and very different days. Keep coffee if you like it. The pattern is what I'm questioning.
To see the pattern you need the real number, and it's usually higher than the one you'd give. The NHS figures are roughly these.
- A mug of instant coffee, about 100 milligrams. Filter and coffee-shop coffee can be more, and a coffee-shop large can be double a mug.
- A mug of tea, black or green, about 75.
- A can of cola, about 40.
- A small can of energy drink, about 80.
- A 50-gram bar of dark chocolate, under 25.
Tuesday comes to 415 milligrams before any chocolate, and over 500 if the half past nine was a large one from the place by the station. You'd have said three.
If you drink tea and not coffee, you've probably been feeling exempt. Six mugs is 450 milligrams, more than four coffees. Tea hides behind its reputation as the drink for shock and comfort, so a person who'd never have a coffee at nine at night has a tea at nine, lies awake at half past eleven, and never connects the two. Everything here applies to you, with smaller doses and more of them. Your taper is easier, because tea steps down naturally, and decaf tea in a mug with milk is hard to tell apart.
"I just love the taste." I believe you. But do you love it at ten past three, instant in a chipped mug, drunk in four gulps at the desk? If switching that cup to decaf feels like being cheated, notice what you'd lose. The taste would still be there. What you'd miss is the pick-up, which is the shortfall being relieved. Keep the coffee you drink for the taste, in the morning, made properly and drunk sitting down. The three o'clock cup is a dose, and a dose can be decaf.
"It's my break." The break is real and you need it. It's made of standing up, the walk, the kettle and four minutes at the window when nobody wants anything. One afternoon this week, take it with a peppermint tea in the mug and see whether it was still a break.
The hardest objection is "I need it because I sleep so badly". For some people that's true. Insomnia, sleep apnoea, pain, a baby and shift work all ruin sleep on their own, and if one of those is yours, the thing itself needs fixing, with a GP if needed. For most readers, though, it runs in a circle. Coffee makes the night shallower, so the morning is a shortfall and the afternoon a slump, and the answer to both is more coffee, which spoils the next night. From inside, that feels exactly like being a bad sleeper who needs coffee. You can't decide to sleep more deeply. You can decide when the last cup is. And if a month at a lower level with an early cut-off leaves you sleeping just as badly and just as tired, your poor sleep has another cause, and that's worth taking to a GP.
Two o'clock
Get the week of notes out and add it up.
Price each item with the NHS figures and be honest about the sizes. Work out a daily total and an average. Then, for each day, find the time of the last caffeine of any kind and set it against your bedtime. Most people find their last dose is later than they thought, because the count stopped at coffee and the afternoon tea never registered.
Then set a cut-off. It's a line, after which you have no caffeine at all. Six hours before bed is the minimum the evidence supports, and I'd go earlier, because your afternoon may hold more than one dose. Two o'clock suits most people. After it, decaf, herbal tea or water. Tea counts. Cola counts.
Move the cut-off first and change nothing else. Leave the morning cups alone. This is the smallest change with the largest effect on the night, and it costs you nothing at seven, where the pick-up feels most needed. Keep the notes going for that week, the times, when you fell asleep and the morning score. In my experience the score often starts to move before the total has come down at all, and that's when the habit stops looking like your nature and starts looking like a schedule.
Once the cut-off has held for a week, bring the total down. This is where people expect to suffer, and most of the suffering is avoidable.
- Decide where you're going first, so the taper has an end. It needn't be zero. One or two cups in the morning, drunk because you like them, is a reasonable place to land. Some people want none, for the pleasure of waking at the level with no cup at all.
- Go slowly. Cut roughly a cup's worth every three or four days and let each step settle. Five hundred milligrams down to two hundred over three weeks will bring a few flat afternoons. Five hundred to nothing on a Monday brings a headache and a bad week, and it will probably fail on Thursday.
- Use decaf and dilution. The afternoon cup has already gone with the cut-off. Next make the morning cups half decaf, then swap the second for tea, then for a weaker tea. Decaf mixed into ground coffee, or one ordinary tea bag among decaf ones, lets you step down without changing the ritual.
- Replace each removed cup with water or a decaf, because some of the headache comes from losing the fluid.
- Expect a few flat days, with the worst around the second day after each step. A painkiller for a headache is fine, and so is a slower step. Treat a flat day as a sign you're adapting. It ends.
- Pick a quiet fortnight, away from a big deadline or a long drive.
- Keep the morning cup till last. Once everything else has gone there's little left to withdraw from, and you may find you want one where you used to need two.
That leaves ten past three, which now has no dose in it.
For the first week or two the fog there is partly withdrawal, and it passes. After that it's the natural dip. Name it. "That's the dip" is a different experience from "I can't function". Take the break anyway, with decaf or water in the mug, or better still go outside for ten minutes, round the block or to the postbox. In my experience daylight and movement do more for that hour than anything in a cup, partly because the dip runs on the body clock and light is what the clock listens to. Have a glass of water, since a surprising share of afternoon fog is mild thirst. Eat a proper lunch, neither skipped nor huge. Put the dull work in the dip and the hard thinking at ten in the morning, which you'll now be awake for. If after a month the afternoon is still a struggle, look at how many hours you're actually in bed, because the coffee may have been hiding a plain lack of sleep.
All of this comes down to one rule. Put it in your own words if you like.
I drink coffee because I like it, before the cut-off, and I don't use it to get through the day.
The moments when the old habit tries to restart are predictable, so decide them in advance. Peter Gollwitzer and Paschal Sheeran pooled ninety-four tests of plans in the form "when X happens, I'll do Y", and found they make people more likely to follow through. Three are worth writing down.
- When I feel the fog at three and head for the kettle, I'll make a decaf and stand at the window for the same four minutes.
- When someone offers me a coffee after the cut-off, I'll say "I'd love a decaf if you've got one, or a tea," and leave it there.
- When I hear myself say "I can't function without it," I'll remember that the fog is the shortfall, and it ends when the shortfall does.
Ten past three, some weeks on
Some days won't fit the rule, and a few need their own.
If you work nights or rotating shifts, the cut-off is six hours before whenever you sleep. Caffeine early in the shift and none in the second half fits the evidence. Shift work damages sleep by itself, and if you're struggling, that's a conversation with a GP or occupational health.
Energy drinks get counted, cut off and tapered like anything else. They're cold and sweet and easy to drink fast, so several go down without notice. No safe amount could be set for children and teenagers, so they aren't for them.
If you have a heart condition, high blood pressure or anxiety, or take a medicine that interacts, follow what you've been told. If nobody has told you anything, ask a pharmacist. If caffeine gives you palpitations, a GP should know.
A single coffee before a long drive when you're tired is the right use of caffeine. It reduces sleepiness for a while. It doesn't make a tired driver safe, so if you're sleepy at the wheel, stop.
At work, have the morning cups with everyone else and a decaf at three, and nobody can tell what's in the mug. At home, tell whoever makes the drinks once. "I'm not having caffeine after two, so make mine a decaf, and I'll keep some in." A deadline week when you decide to have the afternoon cup, knowing the price, is a choice you made. The habit is the week after, when the deadline's gone and the three o'clock cup hasn't.
You'll slip. A late espresso at a birthday dinner, a bad night, a Wednesday that feels like the old days. Go back to the cut-off the next day. There's no need to restart the taper, because one cup doesn't undo three weeks of adapting.
Here's how it might look.
It's ten past three on a Tuesday, some weeks on. The paragraph in front of you is a bit dull, and there's the start of the dip, a slight heaviness you can name now. You put the kettle on, make a decaf, and stand at the window for four minutes watching a van try to reverse into a space. You go back and the paragraph is fine. It was the paragraph.
You woke at twenty past six, before the alarm, which still surprises you, and the fog wasn't there. Coffee at seven from the good beans, drunk sitting down. A second at nine, and that's the count. The drive home is fine without a cup. At half past nine you notice you're drowsy, which used to happen never, and at half past ten you go up and fall asleep without knowing when it happened.
It wasn't always like that. The first week, with only the cut-off moved, the drive home was flat and you resented it. The second week you halved the morning cups, had two days of headache and a foul temper, and nearly stopped. In the third week the morning score in the notebook went from fives to sevens and stayed there, and that was the week you stopped calling yourself a tired person.
You still drink coffee, and you like it more, because it isn't doing a job any more.
If you've read this far and changed nothing, try one thing this week. Keep everything as it is except the last dose. Move it to before two o'clock, and see what the mornings do.