Caffeine half-life calculator
How long does caffeine stay in your system?
By Paul Tanswell · Data checked 2026-10-02
Caffeine's half-life in healthy adults is about 4 hours on average, with a range of about 2 to 8 hours (EFSA, 2015). Take the 108 mg median espresso from our measurements: on the average, about 54 mg is left after 4 hours, 27 mg after 8 and 7 mg after 16. On an 8 hour half-life, 27 mg is still there 16 hours later. The calculator draws that whole range for your day, adjusted only for the three factors with measured effects.
Example, edit it: the figures below update as you change anything
Drinks today
- Espresso, single shot30 ml (1 fl oz)median of 124 measured servings; measured 51 to 322 mg
- Brewed (drip) coffee237 ml (8 fl oz)USDA reference figure; measured 21 to 208 mg
You
Caffeine remaining through the day
Remaining caffeine in milligrams against time of day, as a band across the half-life range with the centre of the range as a line, each drink marked, its peak window hatched, EFSA's 100 mg sleep note as a dotted line and bedtime as a vertical line.
Caffeine remaining through the day
Remaining caffeine in milligrams against time of day, as a band across the half-life range with the centre of the range as a line, each drink marked, its peak window hatched, EFSA's 100 mg sleep note as a dotted line and bedtime as a vertical line.
- Half-life range, with the 4 hour line as the centre
- A drink
- Peak window, 30 to 120 minutes after drinking
- 100 mg: may affect sleep in some adults, close to bedtime (EFSA)
Your last drink peaks between 14:30 and 16:00, going by EFSA's 30 to 120 minutes to peak blood level.
At 23:00, about 5 to 73 mg is still on board, about 28 mg on the 4 hour half-life.
The 4 hour line stays under 100 mg from about 15:40; across the 2 to 8 hour range, from about 14:15 to about 19:25.
Today's total is 203 mg, below EFSA's daily figure of 400 mg, which EFSA states for a 70 kg adult. Your largest serving is 108 mg, below its single-dose figure of 200 mg.
Simplifications: each drink is counted in full at the time drunk, although blood levels actually peak 30 to 120 minutes later (the hatched window); one half-life range covers everyone, moved only by the three measured modifiers; smoking and the pill multiply together, which assumes they act independently; there is no tolerance term; EFSA's figures are scaled by weight over 70 kg. This page summarises published research and is not medical advice.
How the half-life works
A half-life is the time for the amount in the blood to halve. It does not matter how much you started with: each half-life removes half of what is left, so the curve flattens rather than hitting zero. EFSA's mean of 4 hours comes with a range of 2 to 8 hours across healthy adults, which is why the chart shows a band.
| Hours after drinking | 2 h half-life | 4 h half-life | 8 h half-life |
|---|---|---|---|
| 2 | 50% | 71% | 84% |
| 4 | 25% | 50% | 71% |
| 8 | 6% | 25% | 50% |
| 12 | 2% | 13% | 35% |
| 16 | 0% | 6% | 25% |
| 24 | 0% | 2% | 13% |
By convention a dose is treated as effectively gone after about five half-lives, about 3% remaining: 10 to 40 hours across that range.
How long does caffeine take to kick in?
Blood levels peak 30 to 120 minutes after an oral dose (EFSA, 2015), which is the hatched window after each drink on the chart. The quickest case we found is caffeine dissolved in water on an empty stomach: 10 young men peaked at about 30 minutes (Blanchard and Sawers, 1983), and even in that small group the half-lives ran from 2.7 to 9.9 hours. None of the primary sources we read gives an onset time earlier than the peak, so the page does not print one.
What changes the half-life
Three factors have a measured, replicated effect on how fast caffeine leaves, and the calculator applies them as multipliers on the whole 2 to 8 hour band. Everything else people ask about is listed below with what the studies found, and leaves the chart alone.
A caveat on the evidence: we found no systematic review or meta-analysis of caffeine's half-life, for the baseline or for any of these factors. The multipliers rest on small pharmacokinetic studies from 1978 to 2005 that agree in direction, one 786-person population study that EFSA reports, and a curated analysis of 141 studies (Grzegorzewski et al., 2021). Several were read from their abstracts; the methodology page lists which.
| Factor | Half-life times | What was measured | Sources |
|---|---|---|---|
| Smoking | 0.65 | Half-life 3.5 h in 13 smokers against 6.0 h in 13 non-smokers (Parsons and Neims 1978), and 190 against 276 minutes in six and six (May 1982). Clearance 1.5 to 1.8 times faster, falling back within three weeks of stopping (Joeres 1988), and 1.22 to 1.72 times by cigarettes a day in the 786-person study EFSA reports. | Parsons and Neims, 1978; May et al., 1982; Joeres et al., 1988; EFSA, 2015 |
| Oral contraceptives | 1.6 | Half-life 7.88 h on the pill against 5.37 h off it (Abernethy and Todd 1985), and 10.7 against 6.2 h (Patwardhan 1980). Clearance about halved after one cycle of a modern 30 microgram pill in 20 women (Balogh 1995). | Abernethy and Todd, 1985; Patwardhan et al., 1980; Balogh et al., 1995 |
| Pregnancy, first trimester | 1.5 | Caffeine clearance a third lower at weeks 14 to 18 than after the birth (Tracy 2005, 25 women), and the liver enzyme's activity down by about a third at weeks 8 to 16 (Tsutsumi 2001, 12 women). No half-life has been measured before week 14. | Tracy et al., 2005; Tsutsumi et al., 2001; Knutti et al., 1982 |
| Pregnancy, second trimester | 2 | Clearance about half the postpartum value at weeks 24 to 28 (Tracy 2005). | Tracy et al., 2005 |
| Pregnancy, third trimester | 3 | Clearance down about two thirds at weeks 36 to 40 (Tracy 2005); half-life 10.5 h against 3 h in the last four weeks (Knutti 1982), and 8.3 h against 3.4 h across pregnancy as a whole (Knutti 1981); clearance 39% of the non-pregnant value at 38 to 40 weeks (Parsons and Pelletier 1982). EFSA puts the half-life at three to four times longer at term, returning to normal between four weeks and four months after the birth. | Tracy et al., 2005; Knutti et al., 1982; Knutti et al., 1981; Parsons and Pelletier, 1982; EFSA, 2015 |
| Factor | What was measured | Sources |
|---|---|---|
| Fluvoxamine (an antidepressant) | Half-life 5 to 31 h on 100 mg a day in eight volunteers (Jeppesen 1996). One small study of one drug. | Jeppesen et al., 1996 |
| Quinolone antibiotics | Ciprofloxacin took the half-life from 5.2 to 8.2 h in ten men (Healy 1989); enoxacin inhibited more, ofloxacin not at all (Harder 1988). The effect is drug-specific. | Healy et al., 1989; Harder et al., 1988 |
| Alcohol | A single large dose took the half-life from 4.0 to 6.0 h (Mitchell 1983); a week at 50 g a day prolonged it by 72% in ten men (George 1986). Two small studies with different designs. | Mitchell et al., 1983; George et al., 1986 |
| Liver cirrhosis | Clearance 0.9 against 1.4 ml/min/kg in eight cirrhotics (Desmond 1980) and 34 against 64 ml/min (Joeres 1988), falling with disease grade (Holstege 1989). Severity-dependent, so no single multiplier fits. | Desmond et al., 1980; Joeres et al., 1988; Holstege et al., 1989 |
| CYP1A2 genotype (rs762551) | No difference between genotypes in 185 non-smokers, and a difference only among smokers (Sachse 1999). The variant changes how far the enzyme can be induced, not the resting half-life. | Sachse et al., 1999 |
| Age, healthy older adults | Elimination not significantly different between eight young and eight elderly men (Blanchard and Sawers 1983). | Blanchard and Sawers, 1983 (young and elderly) |
| Body weight | In a curated analysis of 141 pharmacokinetic studies, obesity had no effect on caffeine clearance (Grzegorzewski 2021). Weight changes the dose per kilogram, not the half-life. | Grzegorzewski et al., 2021 |
| Newborns and infants | Half-life 50 to 103 hours at birth, about 14 hours at three to four months and two to three hours by six months (EFSA 2015). Outside this page's scope. | EFSA, 2015 |
| Exercise | One study of twelve found a shorter half-life during an hour of moderate exercise, but the volume of distribution fell too (Collomp 1991). Not replicated. | Collomp et al., 1991 |
| Menstrual cycle phase | No significant difference across the cycle in ten women (Kamimori 1999); slower clearance but an unchanged half-life in the luteal phase in another ten (Lane 1992). | Kamimori et al., 1999; Lane et al., 1992 |
How much caffeine is too much?
EFSA's 2015 opinion found that single doses up to 200 mg do not give rise to safety concerns for the general healthy adult population, and that habitual intake up to 400 mg a day does not for healthy adults who are not pregnant. It states both for a 70 kg adult and gives about 3 and 5.7 mg per kg as the equivalents; the calculator scales the stated figures by your weight over 70 kg, which reproduces them exactly at 70 kg. In pregnancy and while breastfeeding EFSA's figure is 200 mg a day, with no per-kg version. For children and adolescents it calls the data "insufficient" and proposes 3 mg per kg a day.
The US FDA has cited 400 mg a day as an amount not generally associated with negative effects for most adults, adding that people vary widely in both sensitivity and how fast they eliminate it. Health Canada's recommended maximums are 400 mg a day for adults, 300 mg for people who are pregnant, planning to be or breastfeeding, and 2.5 mg per kg a day for children and adolescents.
EFSA states that its conclusions do not apply to people selected on the basis of a disease condition or to groups with distinct vulnerabilities due to genetic predisposition or other conditions, which may require individual advice.
This page summarises published research and is not medical advice. Sources: EFSA, 2015; FDA, 2024; Health Canada, 2025.
Caffeine and sleep
EFSA's sleep statement, the dotted line on the chart, is that single doses of 100 mg (about 1.4 mg per kg for a 70 kg adult) may increase sleep latency and reduce sleep duration in some adult individuals, particularly when consumed close to bedtime. It is a note about a dose taken near bedtime, so comparing the modelled amount still on board with it is our analogy, not EFSA's finding.
A 24-study systematic review and meta-analysis (Gardiner et al., 2023) found that caffeine shortened total sleep time by 45.3 minutes on average (95% CI 29 to 61.5), lengthened the time to fall asleep by 9.1 minutes and cut sleep efficiency by 7 percent. An earlier systematic review without pooled figures (Clark and Landolt, 2017) found the same direction of effect: longer time to fall asleep, less total sleep and less deep sleep. The Gardiner model put the point at which a 107 mg coffee stopped significantly reducing total sleep time at 8.8 hours before bed, and 13.2 hours for a 217.5 mg pre-workout dose, with no cut-off found for a 47 mg black tea. The authors call those a starting point: the cut-offs mark where a confidence interval stops crossing zero, the dose and timing relationship held only for total sleep time, and the studies covered healthy adults with low to moderate habits.
The one trial that varied timing directly gave 12 people 400 mg at bedtime, 3 hours before and 6 hours before (Drake et al., 2013), measured with a headband EEG in a trial funded by the device's maker; even at 6 hours, measured total sleep fell from 7.68 to 6.5 hours. It tested one dose only. The widely repeated claim that "100 mg is fine up to four hours before bed" is not in that paper, nor in any source we read; no primary source prints an hours figure for a 100 mg dose.
Questions
How long does caffeine last?
Its half-life in healthy adults averages about 4 hours, with a range of about 2 to 8 hours (EFSA, 2015). So half of a dose is gone after about 4 hours, a quarter is left after 8, and about 6% after 16 hours; on the slow end of the range a quarter is still there 16 hours later. Smoking shortens the half-life and oral contraceptives and pregnancy lengthen it; the calculator above applies those.
How long until caffeine is completely out of your system?
The usual convention is about five half-lives, when about 3% of a dose remains: roughly 10 to 40 hours across EFSA's 2 to 8 hour range, and about 20 hours at the average. A small amount lingers longer than that in anyone.
How long does it take for caffeine to kick in?
Blood levels peak 30 to 120 minutes after drinking (EFSA, 2015). In one bioavailability study, 10 young men drinking caffeine in water on an empty stomach peaked at about 30 minutes (Blanchard and Sawers, 1983).
Does caffeine last longer in pregnancy?
Yes, by a measured and growing amount. Caffeine clearance was 32.8, 48.1 and 65.2 percent lower at weeks 14 to 18, 24 to 28 and 36 to 40 than after the birth (Tracy et al., 2005). EFSA puts the half-life at 3 to 4 times longer at term. The calculator applies 1.5, 2 and 3 times by trimester.
Does decaf count?
A little. The FDA puts decaf at 2 to 15 mg per 8 fl oz (237 ml) cup, and the one decaf filter coffee in our dataset measured 5 mg per cup. The calculator's drink list includes it at 5 mg.
Does a heavier person clear caffeine faster?
Not in the integrated data. Obesity had no effect on clearance across 141 pharmacokinetic studies (Grzegorzewski et al., 2021). Body weight changes the dose per kilogram, and EFSA states its figures for a 70 kg adult with per-kg equivalents, not how fast the dose leaves.
How this calculator works
Each drink decays as its milligrams times one half, raised to the hours since drinking divided by the half-life, and the drinks add up. The curve is drawn three times, for EFSA's 2, 4 and 8 hour half-lives, each multiplied by the factors you tick. The drink figures are the pooled medians and reference figures from the caffeine study; the constants and their sources are listed on the methodology page.
How much caffeine is in a cup of coffee · Espresso caffeine, shot by shot · Methodology and every source