Why caffeine withdrawal gives you a headache

Published 14 September 2026 · 6 min read

By Loutra · Sources (10)

It isn't tiredness. It's a blood vessel letting go.

You changed something yesterday — a weekend, a trip, a decision. This morning there was no coffee, or coffee three hours later than usual. By lunchtime there's a dull ache with nothing to pin it on: not the noise, not the screen, not an especially short night. You put it down to being tired anyway.

The starting mechanism is straightforward. Caffeine is a competitive antagonist at adenosine receptors — adenosine being the molecule that signals it's time to slow down. Caffeine sits in those receptors without triggering anything, and the signal stops arriving.

Take it daily and your brain compensates in the simplest way available: it makes more receptors. The Caffeine Withdrawal chapter of StatPearls puts it like this: "Chronic, high-level caffeine intake can induce upregulation of adenosine receptors in the CNS, increasing sensitivity to normal adenosine signaling and resulting in clinically significant withdrawal symptoms."

The headache, though, comes from a second effect, and this one is vascular. Merideth Addicott and colleagues scanned 45 volunteers (25 women, aged 18 to 50) after a 250 mg dose of caffeine or a placebo. Caffeine reduced grey matter cerebral blood flow by 27%, down to roughly 60 ml per 100 g of tissue per minute.

One detail in the same paper is worth reading twice: the drop was larger in people who had just abstained (33%) than in people on their usual daily dose (20%). The blood vessel never fully gets used to it.

The reverse has been measured directly too. Stacey Sigmon and colleagues maintained 16 regular caffeine users on 400 mg a day for at least fourteen days, then on placebo for at least fourteen days, double-blind. Twenty-four hours after the switch to placebo, blood flow velocity had risen in both the middle and the anterior cerebral arteries — while ratings of "tired", "weary" and "sluggish" went up and ratings of "energetic" and "lively" went down. The authors present this as "the most rigorous demonstration to date of physiological effects of caffeine withdrawal".

⚠️ What neither study demonstrates is that the rising flow produces the pain itself. Addicott states it in the discussion, leaning on earlier work: "withdrawal headaches among caffeine users correspond with increased CBF following 20-24 hours of abstention from caffeine." StatPearls repeats the same reasoning ("Increased cerebral blood flow from loss of vasoconstriction also contributes to headache"). It's the field's standard explanation, not a result obtained inside either experiment.

The timetable doesn't depend on you

In 2004, Laura Juliano and Roland Griffiths reviewed everything published on the subject: 57 experimental studies and 9 surveys. Of 49 symptom categories identified, ten met their validity criteria — headache, fatigue, decreased energy, decreased alertness, drowsiness, decreased contentedness, depressed mood, difficulty concentrating, irritability, and feeling foggy.

The numbers that matter: headache hit 50% of participants, and 13% reported clinically significant distress or functional impairment. The timetable is strikingly consistent from one study to the next.

• 12 to 24 hSymptoms begin, counted from the last dose.
• 20 to 51 hPeak intensity. The worst day and a half.
• 2 to 9 daysTotal duration, before things return to normal.

The most telling piece of evidence is still the study Kenneth Silverman and his team published in the New England Journal of Medicine in 1992. Sixty-two adults, low to moderate intake — 235 mg a day on average, about two and a half cups. Two two-day periods on a caffeine-free diet, during which they received, double-blind, either capsules containing their usual dose or a placebo. Nobody, participants or experimenters, knew which was which.

The result: 52% had a moderate or severe headache during the placebo period, against 2% at baseline and 6% on caffeine. Thirteen per cent took medication the protocol hadn't authorised during the placebo period, against 2% on caffeine. None of that can be put down to what they expected: they didn't know.

One cup a day is enough

This is the counter-intuitive part. Juliano and Griffiths are explicit: incidence and severity rise with daily dose, but "abstinence from doses as low as 100 mg/day produced symptoms".

What does 100 mg look like? The NHS gives the equivalents on its pregnancy nutrition page: "100mg in a cup of instant coffee", "75mg in a cup of tea", "40mg in a can of cola", "80mg in a 250ml can of energy drink". One cup. You don't have to be someone who runs on coffee to collect the headache.

That doesn't mean dose is irrelevant. StatPearls considers heavy habitual users, "particularly those consuming more than 300 mg/day", to be at increased risk, and adds that "individuals with chronic migraines may experience intense attacks following abrupt cessation". And there's a detail in the DSM-5 criteria that surprises most people: the trigger is "abrupt cessation of or reduction in" caffeine use. Going from four coffees to one counts.

What the NHS actually says

The NHS is the public health service that goes furthest on this. Its tension headache page — reviewed on 1 August 2025 — lists three common causes: "stress", "sleep problems", "caffeine". And in the list of things not to do: "do not have drinks with caffeine in them like tea, coffee or cola - or try to cut down on them."

⚠️ Worth flagging, because it's a limit of the page rather than a reading of it: the NHS writes "caffeine" without saying whether it means having it or missing it. Both cause headaches. The distinction lives in the literature, not on that page.

On how to go about cutting down, it's clearer. On its tiredness page it writes that "if you do want to cut caffeine out of your diet completely then the charity The Sleep Charity recommends you reduce your intake gradually".

The withdrawal is a diagnosis. Caffeine addiction isn't.

Here is the inversion that makes this subject different from every other one on this blog.

Sarah Meredith, Laura Juliano, John Hughes and Roland Griffiths sum it up in two sentences in their 2013 review: "Both the ICD-10 and the DSM-5 recognize a clinical diagnosis of Caffeine Withdrawal." And just before: "Although neither the DSM-IV nor the DSM-5 officially recognizes these disorders applied to caffeine, the DSM-5 recognizes Caffeine Use Disorder as a condition for further study." The withdrawal is in the manual; the addiction is in the appendix of things worth looking into.

The withdrawal criteria are also stricter than people assume. They require prolonged daily use, then abrupt cessation or reduction, followed within 24 hours by three or more symptoms from: headache; marked fatigue or drowsiness; dysphoric mood, depressed mood or irritability; difficulty concentrating; flu-like symptoms (nausea, vomiting, or muscle pain and stiffness). The headache on its own is not the diagnosis — it's its most frequent symptom.

Which is the thing to hold on to if you're counting days on something other than coffee: this headache is not a score on how dependent you are, and not proof that you're hooked. It's a pharmacological symptom that shows up at one cup a day, in one person out of two, on a clock.

If you've just quit smoking, your coffee changed dose on its own

Smoking speeds up caffeine clearance — it's one of the factors StatPearls lists as shifting caffeine's half-life, alongside pregnancy, obesity, oral contraceptive use and altitude. The size of the effect has been measured: in smokers who quit and carry on with their usual coffee, concentrations reach 203% of baseline three weeks after quitting (Swanson 1997). Same amount of coffee, double the dose. If you stopped smoking recently and you're sleeping badly, start there — that's the subject of the article on quitting and insomnia.

Don't conclude that you should cut everything at once, though. The same trial split 162 smokers between keeping their caffeine and giving it up too: no difference in other withdrawal symptoms and none in success at 16 days, 6 months or 12 months — and the caffeine abstainers reported more fatigue in the first 3 days. Stacking two withdrawals has never shown a single benefit.

Step down rather than cut out

On what to do, the sources agree and they're brief. StatPearls: "Gradual tapering of caffeine over several days to weeks is preferred over abrupt cessation", with "adequate hydration and rest, particularly during the first 1 to 3 days". The NHS points the same way. The prognosis is unambiguous: "excellent", "self-limited and nonlife-threatening".

⚠️ None of the sources consulted cites a randomised trial comparing tapering against stopping outright for preventing the headache. It's a recommendation resting on pharmacology, not a measured result. That doesn't make it wrong; it explains why nobody gives you a precise schedule.

What you can take away instead of a method: the headache is not a verdict. It starts at a predictable hour, peaks around day two, and leaves. While it lasts, nothing is failing and nothing is being "flushed out" — a blood vessel is returning to the width it had before. And if your headache doesn't look like that at all, caffeine isn't the explanation: the NHS lists headaches "several times a week or … severe", headaches that painkillers don't touch, and feeling sick with light or noise painful as reasons to see a GP — and a headache "that came on suddenly and is extremely painful" as a reason to call 999. A blog post examines nobody.

Questions people ask

Loutra is free on the App Store

An otter that’s counting on you, a daily check-in, and a breathing round for the minutes that are hard.