It's the third mug of the day and the spreadsheet finally behaves, so you keep the coffee arriving until the afternoon turns grey. Then 10:40 that night finds you staring at the ceiling, body wrecked, brain still at the office, and the midnight search question lands in your head: when should you stop drinking coffee. The honest answer is arithmetic rather than folklore, and it only needs three numbers, caffeine's half-life, the size of your last dose and the time you go to bed.
Table of Contents
- The floor is six hours: a fixed 400 mg dose disrupted sleep even when taken 6 hours before bed in a randomised trial, so a 10 p.m. bedtime means nothing caffeinated after 4 p.m.
- The comfortable target is 10 hours: halving and halving again, a morning coffee still leaves real caffeine on board at dinner, which is why early-evening "I feel fine" lies about your brainstem.
- 400 mg is a daily ceiling, not a goal: per the FDA that's roughly two to three 12-ounce cups for most adults, and sensitivity varies enormously between people.
- Decaf isn't zero: an 8-ounce cup holds 2 to 15 mg, which matters mostly if you react strongly or count every late-day milligram. If a magnesium wind-down is already in your plan, the magnesium glycinate vs citrate breakdown covers which form fits evenings.
- Cut back gradually: stopping cold after daily use invites a weekend of headaches and fog. A slow taper over one to two weeks avoids the worst of it.
1. The Half-Life, Not the Feeling
Caffeine clears your body on a clock, and the clock is stubbornly slow. Most healthy adults feel the peak within an hour or so, but MedlinePlus puts the window where you still feel effects at four to six hours per dose. That is the feeling fading. The chemistry takes considerably longer. The half-life of caffeine sits around five hours for a typical adult, which means that five hours after your cup, half the dose is still circulating. One half-life later, a quarter remains. A quarter of a big café latte at bedtime is nothing to hand your brainstem.
The half-life number itself bends with people and habits. Smoking speeds caffeine clearance, while pregnancy, oral contraceptives and certain medications slow it down, sometimes dramatically. Two coworkers can drink the same 3 p.m. cortado and end the night with wildly different amounts left in their systems. There's no shame in being the slow one; there's just a curfew that needs setting on the slow schedule.
The 3 p.m. crash you're chasing often starts at the water cooler long before the coffee machine, though. If your afternoon is flat no matter what you drink, the weight-based math on how much water you actually need is worth running once before you add a fourth cup to the day.
2. The Caffeine Curfew Math, Run on Your Bedtime
Here's the whole method, in one paragraph. Write down your usual bedtime. Subtract six hours for the hard floor. Subtract ten for the generous cutoff you'd aim at while sleep is shaky. Those two numbers are your curfew, and the coffee math is otherwise complete.
Worked example. Bedtime 10:30 p.m. puts the absolute floor at 4:30 p.m. and the working cutoff at 12:30 p.m. That "harmless" after-dinner espresso at 7 p.m.? It arrives at lights-out carrying more than half its dose, roughly 30 to 40 mg of a single shot still active, and it keeps adding minute-counts to the time it takes you to fall asleep.
The FDA's consumer guidance page keeps the numbers honest about the drinks that hide in plain sight: “Decaffeinated coffees and teas have less caffeine than their regular counterparts, but still contain some caffeine.” That sentence is doing quiet work. Your 6 p.m. "decaf treat" isn't a curfew violation. It's a rounding error. The fourth full-strength drip at 2 p.m. is the event.
3. What Six Hours Actually Buys You
Sleep hygiene advice circulates for decades before anyone randomises it. For caffeine, someone finally did. Christine Drake's team at Henry Ford Hospital gave eighteen good sleepers 400 mg of caffeine, a big mug's worth, at 0, 3 and 6 hours before bed on separate nights in a controlled crossover design, with a validated sleep monitor doing the reporting so nobody's memory was on trial.
| Finding | Size of the effect | Study, year |
|---|---|---|
| 400 mg even 6 h before bed | 1.1 to 1.2 hours less total sleep; sleep-onset latency up 17 min at the 3 h mark | Drake et al., J Clin Sleep Med, 2013 |
| Pooled across 22 controlled crossover trials (~1,000 people) | Total sleep down ~35 min, sleep efficiency down 4.7%, falling asleep slower by ~8 min | Chang et al. meta-analysis, Sleep Medicine, 2025 |
| Deep (slow-wave) sleep | Shrinks with higher doses; REM share barely moves | Chang et al. meta-analysis, 2025 |
| Subjective "I slept fine" | Common and unreliable; portable monitors found damage people didn't report | Drake et al., 2013 |
| Single dose vs daily habit | Habitual drinkers show both blunted and persistent disruption; tolerance is partial | Chang et al., 2025 |
Drake's own conclusion line is about as direct as papers get: “The magnitude of reduction in total sleep time suggests that caffeine taken 6 hours before bedtime has important disruptive effects on sleep and provides empirical support for sleep hygiene recommendations to refrain from substantial caffeine use for a minimum of 6 hours prior to bedtime.” An hour of stolen sleep, per night, from a drink taken while the sun was still up.
4. Your Real Tolerance Is a Moving Number
"My roommate sleeps like a rock after an espresso martini." Sure, that person exists. Sleep studies keep finding exactly this split, and the pooled numbers behind the table above say who actually takes the damage. “Both high- and low-dose caffeine reduced total sleep time and sleep efficiency, and increased sleep onset latency, while high-dose caffeine additionally reduced SWS%.” Read that carefully: even a modest dose buys you a shorter night. The heavy habit buys you a shorter night minus the deep sleep that runs on it.
Then there's the part tolerance can't fix. Deep sleep is where growth hormone does its slow work and where your immune system files its nightly paperwork; the meta-analysis found that architecture shrinking with higher doses even while REM looked untouched. You can get used to falling asleep wired. You can't get used to the wiring. That's the honest summary of what late caffeine does.
Some medications quietly change caffeine's speed, too, including a few antibiotics and asthma drugs. If your pharmacist has ever warned you about dairy or sunlight, this is the same idea: one molecule, many metabolisms. Any medication list longer than a receipt is a good reason to set the curfew on the ten-hour side.
5. A Week-Long Reset Without the Headaches
Ordered, timed, and deliberately boring. Run it for seven days and you'll know your actual cutoff, which is worth more than any internet rule of thumb.
- Day 1, evening: write your true bedtime and calculate both cutoffs from section 2. Photograph the clock times on your phone; that photo is the protocol.
- Days 1-2: change nothing except moving all caffeine behind the six-hour floor. Keep total cups where they are so withdrawal doesn't join the party.
- Day 3: push the last cup behind the ten-hour cutoff. After that point, cold brew in the fridge counts as dessert.
- Days 4-5: drop total daily milligrams by about a quarter, mostly by shrinking the biggest cup rather than killing a small one. Smaller, earlier, fewer.
- Day 5 onward: give the freed evening something to do. A wind-down that replaces the ritual matters more than the subtraction itself, and this evening routine for better sleep is built for exactly that slot.
- Day 6: test the floor once, properly. One normal 2 p.m. coffee, bedtime unchanged, and watch what your morning does.
- Day 7, decision: keep whatever cutoff the week earned. If you fell asleep fast and stayed down, you found it. If not, the curfew moves earlier again, or the problem was never the coffee.
6. What's Actually in the Cup
The curfew only works if the milligrams are real. Numbers below are the FDA's consumer-update ranges, published 2024 and last checked against their page that year.
| Drink or food | Typical serving | Caffeine |
|---|---|---|
| Brewed coffee | 8 fl oz | 80 to 100 mg |
| Soft drink | 12 fl oz can | 30 to 40 mg |
| Black or green tea | 8 fl oz | 30 to 50 mg |
| Decaf coffee | 8 fl oz | 2 to 15 mg |
| "Energy drinks" | 16 fl oz | 54 to 328 mg |
| Energy bars, gum, pre-workout, some pain relievers | varies | undeclared unless added directly |
Two rows deserve notes. The energy-drink span isn't a typo: one brand's 16-ounce can can legally carry six times another's, and some of that total hides in guarana, which the FDA counts as caffeine but a casual label-reader often doesn't. The last row is why the "I barely drink coffee" crowd sometimes has the worst sleep in the house. A pre-workout scoop before a 6 p.m. lift is a caffeine curfew violation with gym branding.
7. The Exceptions That Flip the Rule
Pregnancy and breastfeeding change the arithmetic enough to warrant their own paragraph. Caffeine crosses the placenta, and MedlinePlus explicitly tells pregnant and breastfeeding readers to check with a provider about limits; the FDA's 400 mg ceiling is written for healthy non-pregnant adults, full stop. Kids and teens sit outside the rule as well. The FDA's page leans on the American Academy of Pediatrics here, and the position is simply that children and adolescents shouldn't be drinking this stuff, where even a modest can lands harder on a 90-pound body than it does on an adult.
If you take any diuretic, a lithium prescription, or an antibiotic that slows drug metabolism, the curfew exists but the ceiling may be lower than you think. Same conversation as every other dose question: with the prescriber, not the internet. And if you've tightened the cutoff for two full weeks and the insomnia hasn't budged, the coffee has probably been paying someone else's rent. Snoring, gasping or morning headaches point somewhere else entirely, and those deserve a clinician's visit rather than one more experiment with coffee.
- Order the "half-caf" without apology. It halves the milligrams while keeping the mug-and-break ritual intact, which is most of what people actually miss.
- Front-load. Two cups before noon plus one after lunch routinely beats one cup spread into evening, same total, opposite nights.
- Read your pre-workout label tonight. Anything past ~150 mg that's taken after 4 p.m. is a hidden curfew break, and "pump formulas" are where the big numbers hide.
- Track fall-asleep time for one week instead of coffee for one week. The bedtime number shows effects that the cups-per-day number hides.
Do not microdose pure caffeine powder, tablets or "energy crystal" products as a coffee replacement; the FDA estimates toxic effects including seizures can appear with roughly 1,200 mg taken fast, and a level teaspoon of pure powder can hold that dose by itself. Anyone with uncontrolled high blood pressure, a diagnosed arrhythmia, an anxiety disorder being actively treated, or a pregnancy should talk to a clinician before keeping current intake levels, let alone increasing them. If you get racing heart, chest tightness, or a spell of feeling faint after late caffeine, skip the taper and the reset week entirely and contact a clinician the same day.
8. Straight Answers
Is 2 p.m. too late, or is that folklore?
Folklore with a decent core. With a 10-hour cutoff, a 2 p.m. last cup is perfect for a 10 p.m. bedtime and late for an 8 p.m. one. Your bedtime decides; the clock on the wall doesn't.
Can I keep a morning coffee if I feel totally fine at night?
Please do, that's the whole point of the six-hour floor rather than a zero-caffeine rule. "Fine" gets one free pass; if the monitors disagreed with how you felt in the studies above, it's worth checking how fast you fall asleep and how many times you'd estimate you wake.
Does decaf count toward the curfew?
Barely. Decaf holds 2 to 15 mg per cup, and that's a rounding error unless you react strongly to traces or you're counting milligrams for a medical reason.
What if the headache won't let me cut back?
That's withdrawal, not weakness, and it's time-limited. Slow the taper to a quarter-cup reduction every other day, keep fluids up, and keep any pain reliever within its label. A headache lasting beyond a fortnight of tapering deserves a real workup.
Does caffeine before exercise still ruin sleep if I work out at 5 a.m.?
The early session clears caffeine from the equation entirely, which is exactly why 5 a.m. people drink it at 4:30 without consequence. The timing rule follows the distance to bedtime, not the distance to the gym.
- U.S. Food and Drug Administration (2024). 400 mg/day framing for most adults, caffeine per drink, decaf content, and concentration-product risks. FDA Consumer Updates. [View Study]
- Drake C, Roehrs T, Shambroom J, Roth T (2013). 400 mg of caffeine at 0, 3 or 6 hours before bed each disrupted measured sleep; supports a minimum six-hour abstention. J Clin Sleep Med. [View Study]
- Chang YH, et al. (2025). Pooled effects of caffeine on objective sleep across 22 controlled crossover trials; age and dose findings. Sleep Medicine. [View Study]
- MedlinePlus (n.d.). How long caffeine's effects last, who should limit it, and withdrawal symptoms. NIH National Library of Medicine, Caffeine in the Diet. [View Study]
How we checked: every figure here was read in the original paper or government page before publication, and the page was edited by our editorial team rather than auto-published.
Read next: Magnesium Glycinate vs Citrate: Sleep or Muscles? · Evening Routine for Better Sleep That Actually Helps
This article is for informational purposes only and is not a substitute for professional medical advice. Always consult a qualified healthcare provider before making changes to your diet, exercise, or health routine.
💬 Comments
Post a Comment