Cutting back
Caffeine tolerance: why coffee stopped working, and how to reset it
Updated 2026-10-03 · Checked against the sources below by Oleksii Y. · 7 min read
Coffee stops feeling like it works because your body adapts to daily caffeine. In one controlled study, people taking high doses every day became fully tolerant to its felt effects, and in a trial of 379 people, regular users got no net alertness boost: going without lowered their alertness and caffeine only restored it. A break may lower your tolerance, but we found no study that measured how long a reset takes. What is measured is the cost: withdrawal typically lasts 2–9 days.
Key takeaways
- Tolerance to caffeine’s felt effects can be complete: in one study it was there after 18 days of high daily doses.
- Regular users develop substantial tolerance to caffeine’s anxiety effect, and may gain no alertness above their normal baseline.
- Tolerance is uneven: the cortisol response shrinks but doesn’t disappear, and it is unclear whether sleep adapts at all.
- We found no study that measured how long a tolerance break takes to work; “7–14 day resets” are not backed by direct evidence.
- A break means withdrawal first: symptoms typically start 12–24 hours after your last caffeine and last 2–9 days.
What caffeine tolerance is, and the study that measured it
Tolerance means the same dose does less than it used to. MedlinePlus lists dependency among the effects of too much caffeine, describing it as needing more caffeine to get the same results.
The clearest evidence comes from a 1992 study by Evans and Griffiths. Thirty-two healthy adults stopped all dietary caffeine, then took either 300 mg of caffeine three times a day (900 mg a day) or a placebo for 18 days. Afterwards, a test dose of caffeine still produced clear subjective effects in the placebo group but not in the group that had taken it daily. The authors described it as the strongest evidence then available that people can become completely tolerant to one of caffeine’s effects on the brain.
Two cautions: 900 mg a day is far above typical intake, and the study does not tell you how many days tolerance takes at one or two cups. So “it takes X days to build tolerance” is not something the evidence can answer for ordinary drinkers.
Does daily coffee still wake you up, or just undo withdrawal?
This is the heart of “coffee doesn’t work anymore”. In a randomized trial by Rogers and colleagues, 379 people who were either non- or low users or medium- to high users went 16 hours without caffeine, then received caffeine or a placebo. In the regular users, the placebo lowered alertness and raised headache, and caffeine merely returned alertness to baseline: no net gain.
A 2005 review by James and Rogers went further, arguing that the performance and mood benefits regular users feel are almost wholly the reversal of withdrawal from short gaps without caffeine, because tolerance develops to the modest effects caffeine has at first.
Other researchers disagree. A review by Smith found caffeine increases alertness and reduces fatigue, especially when alertness is low, such as at night. In 102 light, non-dependent users, Childs and de Wit found that caffeine produced feelings of arousal and sharpened performance on a vigilance task, which they took as evidence that caffeine has real effects without withdrawal. The honest summary: for daily drinkers, a large part of the morning lift may be withdrawal relief, but the debate isn’t settled. Our guide to why coffee can make you tired follows this thread.
Which caffeine effects fade with tolerance, and which may not
Tolerance is not all-or-nothing. Some effects fade strongly, others only partly, and for sleep nobody knows yet. That matters, because feeling nothing from your coffee is not proof that it does nothing.
Sleep is the effect to watch. In a 2025 trial, the authors concluded that whether habitual users become tolerant to caffeine’s sleep disruption remains unclear, and they found people may have difficulty perceiving how caffeine affects their sleep. If your coffee no longer wakes you up but you sleep poorly, see our guide to coffee before bed and check your timing with the caffeine cutoff calculator.
| Effect | What the research found | Study |
|---|---|---|
| Feeling the “buzz” | Complete tolerance after 18 days at 900 mg a day | Evans & Griffiths 1992 |
| Anxiety and jitters | Substantial tolerance with frequent use, even in genetically susceptible people | Rogers et al. 2010 |
| Alertness | No net gain above baseline in regular users (debated) | Rogers et al. 2010; James & Rogers 2005 |
| Cortisol (a stress hormone) | Response reduced but not eliminated in daily users | Lovallo et al. 2005 |
| Sleep disruption | Unclear whether tolerance develops | Gardiner et al. 2025 |
How to reset caffeine tolerance: what is known and what isn’t
The idea behind a tolerance break is simple: if daily use dulls caffeine’s effects, time without it should restore them. That is plausible, and it fits the finding that non- and low users in the Rogers trial felt more anxious after caffeine than regular users did. It is an inference, though: that trial compared different people, not the same people before and after a break. Don’t expect a big alertness payoff either: in the same trial, caffeine did not raise alertness in the non- and low users, although Childs and de Wit did see effects in light users. We found no study that measured how long a break needs to be, so confident “7-day” or “14-day reset” claims are not backed by direct evidence.
What has been measured is the cost of stopping. A review of 66 studies found withdrawal symptoms typically start 12 to 24 hours after the last caffeine, peak at 20 to 51 hours and last 2 to 9 days, with headache in about half of people. Whatever break you choose, the first days will likely be the hardest; our caffeine withdrawal timeline shows what to expect.
If you would rather not stop completely, the FDA advises cutting back gradually. A 2016 trial helped heavy users reduce their intake over 5 weeks with lasting results, though it did not test tolerance itself.
- Full break: stop all caffeine, including hidden sources, and expect the typical 2–9 day withdrawal course.
- Step down: cut back gradually, as the FDA and MedlinePlus advise, for example by swapping some cups for decaf or half-caf.
- Coming back: start with a small dose. Light users in one study felt more anxious at higher doses, and your response after a break may be stronger than you remember.
Other reasons your coffee seems weaker than it used to
Sometimes the coffee changed, not you. Caffeine in coffee depends on the beans, how it is made and how it is prepared, according to EFSA. In lab tests of US coffee-shop coffees, caffeine ranged from 58 to 259 mg per serving, and the same drink from the same shop varied from 259 to 564 mg over six days. A new café, roast or brewer can change your dose more than you would guess. Our caffeine chart shows the ranges.
Drinking more is not the answer either. Tolerance to the buzz doesn’t raise the amount that is sensible for your body: the FDA’s figure for most adults is still 400 mg a day, an amount it says is not generally associated with negative effects, and our guide to how much caffeine is too much explains the limits.
Finally, if you are exhausted even with plenty of caffeine, tolerance may not be the main story. Persistent tiredness has many causes that coffee can’t fix, and it is worth seeing a doctor.
Common questions
How long does it take to reset caffeine tolerance?
We found no study that measured it, so any precise number is a guess. What is known is the withdrawal you go through first: symptoms typically start 12–24 hours after your last caffeine, peak at 20–51 hours and last 2 to 9 days, according to a review of 66 studies.
Can you lower caffeine tolerance without quitting completely?
Possibly, but it hasn’t been tested directly. In one trial, non- and low users felt more anxious after caffeine than regular users, which suggests a smaller habit means a stronger response, at least for that effect. Cutting back gradually, as the FDA advises, is a reasonable way to try it while limiting withdrawal symptoms.
Why does coffee not wake me up anymore but still keep me up at night?
Tolerance to caffeine’s felt effects can be complete, but researchers say it is unclear whether regular users become tolerant to its effect on sleep. A 2025 trial also found people may not accurately perceive how caffeine affects their sleep. Feeling nothing during the day doesn’t mean your night is unaffected.
Is caffeine tolerance the same as caffeine addiction?
Not quite. Tolerance means the same dose does less. Caffeine withdrawal is a recognized diagnosis in the DSM-5, but caffeine use disorder is listed only as a condition for further study. Needing coffee to feel normal is common and is not, by itself, a diagnosis.
General information, not medical advice
If you feel unwell after caffeine — for example chest pain, a fast or irregular heartbeat that does not settle, fainting, confusion or trouble breathing — get urgent medical help. If you are pregnant, take medication or have a heart, liver or anxiety condition, ask a doctor or pharmacist what amount is right for you.
Sources
- Evans SM, Griffiths RR. (1992). Caffeine tolerance and choice in humans. Psychopharmacology, 108(1–2), 51–59. PMID 1410146 ↗
- Rogers PJ, Hohoff C, Heatherley SV, et al. (2010). Association of the anxiogenic and alerting effects of caffeine with ADORA2A and ADORA1 polymorphisms and habitual level of caffeine consumption. Neuropsychopharmacology, 35(9), 1973–1983. PMID 20520601 ↗
- James JE, Rogers PJ. (2005). Effects of caffeine on performance and mood: withdrawal reversal is the most plausible explanation. Psychopharmacology, 182(1), 1–8. PMID 16001109 ↗
- Smith A. (2002). Effects of caffeine on human behavior. Food and Chemical Toxicology, 40(9), 1243–1255. PMID 12204388 ↗
- Childs E, de Wit H. (2006). Subjective, behavioral, and physiological effects of acute caffeine in light, nondependent caffeine users. Psychopharmacology, 185(4), 514–523. PMID 16541243 ↗
- Childs E, Hohoff C, Deckert J, et al. (2008). Association between ADORA2A and DRD2 polymorphisms and caffeine-induced anxiety. Neuropsychopharmacology, 33(12), 2791–2800. PMID 18305461 ↗
- Lovallo WR, Whitsett TL, al'Absi M, et al. (2005). Caffeine stimulation of cortisol secretion across the waking hours in relation to caffeine intake levels. Psychosomatic Medicine, 67(5), 734–739. PMID 16204431 ↗
- Gardiner CL, Weakley J, Burke LM, et al. (2025). Dose and timing effects of caffeine on subsequent sleep: a randomized clinical crossover trial. Sleep, 48(4), zsae230. doi:10.1093/sleep/zsae230 ↗
- Juliano LM, Griffiths RR. (2004). A critical review of caffeine withdrawal: empirical validation of symptoms and signs, incidence, severity, and associated features. Psychopharmacology, 176(1), 1–29. PMID 15448977 ↗
- MedlinePlus (U.S. National Library of Medicine) (2021). Caffeine: MedlinePlus. MedlinePlus, U.S. National Library of Medicine. medlineplus.gov ↗
- U.S. Food and Drug Administration (2024). Spilling the Beans: How Much Caffeine is Too Much? FDA Consumer Updates. fda.gov ↗
- Evatt DP, Juliano LM, Griffiths RR. (2016). A brief manualized treatment for problematic caffeine use: A randomized control trial. Journal of Consulting and Clinical Psychology, 84(2), 113–121. PMID 26501499 ↗
- McCusker RR, Goldberger BA, Cone EJ. (2003). Caffeine content of specialty coffees. Journal of Analytical Toxicology, 27(7), 520–522. PMID 14607010 ↗
- European Food Safety Authority (2015). EFSA explains risk assessment: Caffeine. European Food Safety Authority (doi:10.2805/618813). efsa.europa.eu (PDF) ↗
- Meredith SE, Juliano LM, Hughes JR, Griffiths RR. (2013). Caffeine Use Disorder: A Comprehensive Review and Research Agenda. Journal of Caffeine Research, 3(3), 114–130. PMID 24761279 ↗
- Rocha Cabrero F, Hamilton RJ. (2025). Caffeine Withdrawal. StatPearls [Internet], StatPearls Publishing (NCBI Bookshelf). PMID 28613541 ↗
An app for better sleep is on the way
We’re working on a mobile app to help you sleep better. It isn’t out yet. Leave your email and we’ll let you know when it launches.