Short answer: symptoms begin 12 to 24 hours after your last dose, peak somewhere between 20 and 51 hours, and are gone within two to nine days. Headache is the signature symptom and shows up in about half of people who stop abruptly. Roughly one in eight gets symptoms bad enough to interfere with a normal day. Those figures come from Juliano and Griffiths' review of 66 studies, and twenty years later nobody has produced better ones.
The timeline is not arbitrary. It follows directly from caffeine's half-life and from what your brain did to cope with the daily dose. Once you see the mechanism, the schedule stops being a mystery and the way to avoid it becomes obvious.
The timeline, hour by hour
Everything below assumes an abrupt stop after regular daily use. If you taper, most of this table does not happen to you, which is the subject of the last section.
| Time since last dose | What is happening | What you notice |
|---|---|---|
| 0–6 h | Caffeine still circulating at useful levels | Nothing. Possibly the mild afternoon dip you always get |
| 6–12 h | Blood levels fall past the point where adenosine is meaningfully blocked | Sluggish, a little foggy. Easy to attribute to a long day |
| 12–24 h | Withdrawal begins. Cerebral blood flow rises as vasoconstriction lifts | Headache starts, usually dull and diffuse. Fatigue, yawning, irritability |
| 20–51 h | Peak. Adenosine signalling is at its most unopposed | Worst headache, marked tiredness, low mood, difficulty concentrating. Flu-like heaviness in some people |
| Days 3–4 | Receptor numbers begin drifting back toward baseline | Headache fading, energy still flat |
| Days 5–9 | Adaptation largely reversed | Most people symptom-free. Slow metabolisers and heavy users are at the long end |
| Week 2 onward | Baseline | Normal alertness without caffeine. Sleep is often measurably better than it was |
Two things about that table are worth internalising. First, the peak is on day two, not day one. People who stop on a Saturday morning often feel fine by Saturday evening and conclude they have got away with it. Sunday is when the headache arrives. Second, the range is wide, two to nine days, and where you fall in it is mostly a function of dose and clearance rate rather than willpower.
Why the headache arrives on a schedule
Caffeine works by sitting in adenosine receptors and blocking them. Adenosine is the molecule that accumulates during the day and makes you sleepy, and it also dilates blood vessels in the brain. Block it and you get two things at once: alertness, and mild cerebral vasoconstriction.
Take caffeine every day and the brain responds the way it responds to any receptor blockade, by making more receptors. That is tolerance, and it is why your third month of coffee does not feel like your first week. It is also the withdrawal mechanism waiting to happen. Remove the caffeine and you have an unusually large number of adenosine receptors with nothing blocking them.
Sigmon and colleagues (2009) measured what this does. Sixteen regular users were held on 400 mg a day, then switched to placebo under double-blind conditions. Withdrawal increased blood flow velocity in the middle and anterior cerebral arteries, shifted the EEG toward the slow theta band associated with low arousal, and produced the expected fatigue and loss of vigour. The vessels that caffeine had been constricting for months dilated, and the increased blood flow is what most people feel as the headache.
The schedule comes from the half-life. At four to six hours, it takes 12 to 24 hours for blood levels to fall far enough that the excess receptors are fully unblocked. That is exactly the onset window. The peak follows as the last traces clear over the next day. The recovery, days three to nine, is the time it takes to downregulate the extra receptors, which is a slower biological process than clearing a drug.
Tolerance and "complete" tolerance were observed on subjective but not physiological measures.
Sigmon, Herning, Better, Cadet & Griffiths, Psychopharmacology, 2009
That line from the same paper deserves a moment. After two weeks of daily caffeine the participants no longer felt much from a dose, but their cerebral blood flow and EEG still responded to it. The subjective effect had gone; the physiology had not. Which is the same pattern as caffeine and sleep: the body keeps responding long after you have stopped noticing.
The symptoms, and how common each one is
Juliano and Griffiths validated the symptom list by requiring that each one appear across multiple controlled studies, not just in self-report surveys. The result is shorter than most online lists.
| Symptom | How well established | Notes |
|---|---|---|
| Headache | Strongest evidence | About 50% of people who stop abruptly. Typically diffuse, throbbing, worse with exertion |
| Fatigue, drowsiness | Strong | The most consistently reported symptom after headache |
| Decreased alertness, difficulty concentrating | Strong | Measurable on cognitive tests, not only in self-report |
| Depressed or irritable mood | Strong | Usually mild, resolves with the rest |
| Flu-like symptoms: nausea, muscle pain, stiffness | Moderate | Less common, more likely at higher habitual doses |
| Anxiety | Weak | Reported in some studies, absent in others |
| "Brain fog" | Not a validated category | Overlaps with the concentration and alertness items above |
Fourteen well-described symptoms cluster into three groups in factor analysis: headache, fatigue and low arousal, and dysphoric mood. If you have all three at once on day two, that is textbook withdrawal, not a coincidental cold.
The DSM-5 lists caffeine withdrawal as a diagnosis, requiring at least three of five symptom groups (headache; fatigue or drowsiness; dysphoric mood or irritability; difficulty concentrating; flu-like symptoms) within 24 hours of stopping, with clinically significant distress or impairment. Roughly 13% of people meet that impairment bar in the studies Juliano and Griffiths reviewed. That is not a small number for something most people manage by accident.
How much caffeine it takes
Less than most people assume, and for less time.
Evans and Griffiths (1999) ran the parametric study. Withdrawal appeared after as little as three days of regular dosing at 300 mg a day, and became more reliable and more severe with 7 to 14 days of exposure. Earlier work from the same group had already shown withdrawal in people whose entire habit was 100 mg a day, which is a single cup of filter coffee or two cups of black tea.
So the person who says "I only have one coffee, I can't be dependent" is very likely wrong on the pharmacology, and the person who picks up a coffee habit on a two-week holiday can expect a headache when they get home. Dose does scale severity: heavier users get worse and longer withdrawal. But the threshold for getting some withdrawal is low enough that nearly every daily user is over it.
| Habitual daily intake | Roughly | What to expect on abrupt stop |
|---|---|---|
| Under 100 mg | One tea, or a small coffee | Often nothing, sometimes a mild day-two headache |
| 100–200 mg | One or two coffees | Headache and fatigue in a substantial minority |
| 200–400 mg | Two to four coffees, or one pre-workout | The classic pattern: day-two headache, several flat days |
| Over 400 mg | Energy drinks on top of coffee, or an all-day habit | Longer and worse, more likely to include flu-like symptoms |
Why yours might be longer
The two-to-nine-day range is wide, and the same factors that stretch caffeine's half-life stretch the withdrawal window with it.
Slow CYP1A2 metabolism. If you clear caffeine slowly, it takes longer to reach the point where withdrawal starts, and the whole curve shifts right. Onset closer to 24 hours than 12; peak on day three rather than day two.
Pregnancy and oestrogen-containing contraceptives slow clearance further, with the same effect. Pregnancy is also the most common reason people stop caffeine abruptly, which is unfortunate timing.
Quitting smoking at the same time. Smoking roughly halves caffeine's half-life. Stop smoking and your usual coffee suddenly lasts twice as long, so people often cut caffeine at the same time, and get both withdrawals stacked.
Dose. Heavier users have more receptor upregulation to reverse. The nine-day end of the range is populated mainly by people who were well above 400 mg a day.
Sleep debt. Caffeine was masking it. When the mask comes off, the tiredness you feel is partly withdrawal and partly the deficit that was always there. The withdrawal component resolves in a week. The deficit does not, until you sleep.
Caffeine withdrawal headache specifically
Since it is the symptom people search for, a few specifics.
It is a vascular headache: dull, often bilateral, frequently described as pressure behind the eyes or across the forehead, worse on standing or exertion. It does not usually come with the aura or one-sidedness of migraine, though people with migraine often find caffeine withdrawal triggers one.
It responds to caffeine within about 30 to 60 minutes, which is diagnostic and also the trap. Many over-the-counter headache tablets contain 30 to 65 mg of caffeine for exactly this reason. Taking one for a caffeine withdrawal headache treats the headache by partially undoing the withdrawal, and resets the clock.
If you want it gone without restarting, plain paracetamol or ibuprofen without added caffeine, water, and time. Day three is usually noticeably better than day two.
How to stop without the timeline
The whole syndrome exists because receptor upregulation takes days to reverse and an abrupt stop exposes it all at once. Reduce slowly enough that the receptors keep pace and there is no gap to fall into.
Sweeney and colleagues (2019) tested a manual-only gradual reduction programme in 36 problematic caffeine users, most of whom met withdrawal criteria at baseline. Participants cut intake by 84% by the end of treatment and were still 75% down twenty weeks later, confirmed by salivary caffeine, without the withdrawal that an abrupt stop would have produced. The method was not sophisticated. It was a schedule.
| Week | Daily target | How |
|---|---|---|
| 1 | Cut 25% | Drop the last coffee of the day. This also helps sleep first |
| 2 | Cut 50% | Replace one remaining coffee with tea (about half the dose) |
| 3 | Cut 75% | Move to tea or a half-strength coffee |
| 4 | Cut 90–100% | Decaf, or stop. Decaf still carries about 2 mg per cup, which is enough to make the last step gentle |
Rules that make it work:
Cut the latest dose first. It does most of the damage to sleep and the least for alertness, so removing it costs almost nothing and pays back immediately.
Do not swap in energy drinks or pre-workout. They are caffeine in a different bottle, usually a larger dose than the coffee they replace.
Keep the ritual, change the contents. Most of what people miss is the cup, the pause and the warmth, none of which is pharmacological. A decaf or a tea at the same time in the same mug covers it.
If you get a headache, you went too fast. Go back to the previous week's dose for a few more days. There is no prize for speed.
What to actually expect afterwards
Two weeks out, most people report that they feel roughly the way they did with caffeine, which is the part that surprises them. That is the "absence of net beneficial effects" in Sigmon's title: after tolerance, the daily dose was largely returning you to the baseline that withdrawal took away each morning. The alertness you were buying was mostly the alertness you had been losing.
What tends to improve is sleep. Caffeine taken six hours before bed reduces objectively measured sleep by over an hour, and most daily users have some evening caffeine on board every night. Removing it is one of the few sleep interventions with an effect size large enough to see on a consumer tracker.
Whether that trade is worth it is a personal decision, and nothing here argues that it must be. But it is worth knowing that the cost of finding out is one difficult weekend if you stop abruptly, or no difficult days at all if you take a month.
Questions people ask
How long does caffeine withdrawal last? Two to nine days after the last dose for most people. Symptoms begin at 12 to 24 hours, peak at 20 to 51 hours, and taper off from day three. Heavy users and slow metabolisers sit at the long end of that range.
When is the worst day of caffeine withdrawal? Day two. The peak window is 20 to 51 hours after the last dose, so a Saturday-morning stop usually produces the worst symptoms on Sunday.
How long does a caffeine withdrawal headache last? It typically starts within a day of stopping, is worst on the second day and fades over the following two to three days. Caffeine relieves it within an hour but restarts the clock; caffeine-free painkillers and time do not.
How much caffeine do you need to get withdrawal? As little as 100 mg a day, roughly one coffee, taken regularly. Withdrawal has been produced experimentally after only three days of daily dosing at 300 mg, and becomes more reliable after one to two weeks.
Can caffeine withdrawal last two weeks? Rarely for the core symptoms, which resolve within nine days in the reviewed studies. Persistent tiredness beyond that is more likely to be sleep debt that caffeine had been covering, or another cause, than ongoing withdrawal.
Does tapering caffeine prevent withdrawal? Largely, yes. Cutting by about a quarter a week lets receptor numbers return to baseline gradually. A randomised trial of gradual reduction achieved an 84% cut in intake without the withdrawal syndrome an abrupt stop produces.
What are the symptoms of caffeine withdrawal? Headache, fatigue or drowsiness, reduced alertness and concentration, low or irritable mood, and in some people flu-like symptoms such as nausea and muscle stiffness. The DSM-5 requires three of these groups within 24 hours of stopping.
Why does caffeine withdrawal cause a headache? Caffeine constricts cerebral blood vessels by blocking adenosine. Daily use increases adenosine receptor numbers, so when caffeine is removed the vessels dilate more than they did before you started. The increased blood flow produces the headache.
Does decaf help with caffeine withdrawal? Slightly. Decaf contains around 2 mg of caffeine per cup, which is pharmacologically small but not zero, and it preserves the ritual. It is a reasonable final step in a taper rather than a treatment for withdrawal on its own.
Will I feel better after quitting caffeine? After the withdrawal window, most people feel about the same during the day as they did with caffeine, because tolerance had already cancelled most of the benefit. Sleep quality is the measure most likely to improve.
This article covers caffeine pharmacology for general information and is not medical advice. A severe or unusual headache, or one accompanied by fever, stiff neck, vision change or confusion, needs medical attention regardless of caffeine.
References
- Juliano, L.M., & Griffiths, R.R. (2004). A critical review of caffeine withdrawal: empirical validation of symptoms and signs, incidence, severity, and associated features. Psychopharmacology, 176(1), 1–29. doi:10.1007/s00213-004-2000-x
- Sigmon, S.C., Herning, R.I., Better, W., Cadet, J.L., & Griffiths, R.R. (2009). Caffeine withdrawal, acute effects, tolerance, and absence of net beneficial effects of chronic administration: cerebral blood flow velocity, quantitative EEG, and subjective effects. Psychopharmacology, 204(4), 573–585. doi:10.1007/s00213-009-1489-4
- Evans, S.M., & Griffiths, R.R. (1999). Caffeine withdrawal: a parametric analysis of caffeine dosing conditions. Journal of Pharmacology and Experimental Therapeutics, 289(1), 285–294. PubMed 10087016
- Sweeney, M.M., Meredith, S.E., Juliano, L.M., Evatt, D.P., & Griffiths, R.R. (2019). A randomized controlled trial of a manual-only treatment for reduction and cessation of problematic caffeine use. Drug and Alcohol Dependence, 195, 45–51. doi:10.1016/j.drugalcdep.2018.10.034
- Rocha Cabrero, F., & Hamilton, R.J. (2025). Caffeine Withdrawal. In StatPearls. StatPearls Publishing. NBK430790
- Nehlig, A. (2018). Interindividual Differences in Caffeine Metabolism and Factors Driving Caffeine Consumption. Pharmacological Reviews, 70(2), 384–411. doi:10.1124/pr.117.014407
The weekly readout
One email each Thursday: what we tested, which claim collapsed under a closer look, and the one number worth paying attention to.