Elmsword
← Sleep

Sleep · Research · 5 min read

Caffeine dose and timing: 400 mg cut deep sleep even 12 hours before bed

In a crossover trial of 23 healthy young men, 100 mg of caffeine had no significant effect on sleep even 4 hours before bed, while 400 mg reduced N3 sleep whether it was taken 12, 8 or 4 hours before bed.

Most advice about caffeine and sleep is about the clock: stop by a certain hour. This trial asked whether the size of the dose matters too. Healthy young men took either 100 mg of caffeine (about one cup of coffee, in the authors' terms) or 400 mg at 12, 8 or 4 hours before bedtime. The 100 mg dose had no significant effect on sleep at any of those times. The 400 mg dose reduced N3 sleep at all three times, and the closer to bedtime it was taken, the more sleep it disrupted. The study did not include women, older adults or people with insomnia. [1]

For a look at caffeine cut-off timing across many earlier studies, see the related article Caffeine and sleep: how early is early enough?

What the researchers did

This was a placebo-controlled, double-blind, randomized crossover trial (registration ACTRN12621001625864). Thirty eligible men volunteered. Three withdrew before the intervention and four were excluded for not following the protocol, so 23 were analyzed: mean age 25.3 ± 5.0 years. To take part, they had to be healthy, aged 18 to 40, good sleepers (Pittsburgh Sleep Quality Index score of 5 or less) and moderate habitual caffeine users (under 300 mg a day). Women were excluded; the authors say accounting for hormonal profiles, including menstrual cycle phase, would have extended the protocol. [1]

Each man completed seven conditions in random order, with a 48-hour washout between them: [1]

  • Placebo: three look-alike capsules with no caffeine.
  • 100 mg: a typical dose, about one cup of coffee in the paper's description, taken 12, 8 or 4 hours before bedtime.
  • 400 mg: a high dose, described as the upper recommended daily limit, taken 12, 8 or 4 hours before bedtime.

On every condition day, participants swallowed one capsule at 12, 8 and 4 hours before their usual bedtime; only one of the three held caffeine, or none on placebo days. Outside the capsules, they took their usual daily caffeine as a capsule within an hour of waking and avoided other caffeine. All testing happened at home. Sleep was recorded with the Somfit, a forehead device the paper describes as validated partial polysomnography, plus a morning sleep diary. Results were estimated with linear mixed models. The authors note that 400 mg is roughly four standard coffees (107 mg per 250 mL), with large variability between products. [1]

Key findings

Estimated difference with 400 mg compared with placebo, from the Results section. * Statistically significant. "Not significant" with a number: the paper reports the estimate and p value. "No significant change": the paper reports significant effects for this outcome only at the other timings. 100 mg: no significant effect on any objective or subjective sleep outcome at any timing. [1]
Outcome (400 mg vs. placebo)12 h before bed8 h before bed4 h before bed
Total sleep time−30.0 min (not significant, p = .060)−28.7 min (not significant, p = .076)−50.6 min*
N3 sleep−20.6 min*−15.3 min*−29.7 min*
Wake after sleep onset (WASO)No significant change+29.1 min*+26.2 min*
Sleep efficiencyNo significant change−6.9%*−9.5%*
Sleep onset latencyNo significant changeNo significant change+14.2 min*
Perceived sleep quality (1–5 scale)No significant changeNo significant change−0.8 units*
Caffeine dose and timing vs. that night's sleepSeven conditions: placebo, and 100 mg or 400 mg caffeine taken 12, 8 or 4 hours before bedtime. Compared with placebo, 400 mg reduced N3 sleep by an estimated 20.6 minutes at 12 hours, 15.3 minutes at 8 hours and 29.7 minutes at 4 hours, all significant. Total sleep time fell by 50.6 minutes at 4 hours (significant); the estimates at 8 hours (28.7 minutes) and 12 hours (30.0 minutes) were not significant. 100 mg had no significant effect.Caffeine dose and timing vs. that night's sleepRandomized, double-blind, placebo-controlled crossover trial, 23 healthy young men, at homeA. Seven conditions per participant, 48-hour washout between themPlaceboall three capsules100 mg caffeineat 12, 8 or 4 h before bed400 mg caffeineat 12, 8 or 4 h before bedCapsules at 12, 8 and 4 h before bed on each condition day · sleep measured with Somfit (partial polysomnography)B. 400 mg vs. placebo: minutes less sleep that nightN3 sleepTotal sleep time12 h before bed20.6 min less*30.0 min less8 h before bed15.3 min less*28.7 min less4 h before bed29.7 min less*50.6 min less** Statistically significant vs. placebo. Total sleep time at 8 and 12 h: not significant (p = .076, p = .060).100 mg: no significant effect on objective or subjective sleep at any of the three timings.Source: Gardiner et al., Sleep 2025, Results (Objective sleep).
Source: Gardiner et al., Sleep 2025, Results (Objective sleep). Bars show estimated minutes of N3 sleep and total sleep time lost with 400 mg compared with placebo. [1]

100 mg had no significant effect, even 4 hours before bed. Neither the device recordings nor the sleep diaries showed a significant difference from placebo at any of the three timings. [1]

“No significant effects were observed on objective sleep outcomes with consumption of the 100 mg dose of caffeine (p > .050).”

Gardiner et al. · Results · full text

400 mg disrupted sleep more the later it was taken. At 4 hours before bed, total sleep time fell by an estimated 50.6 minutes, sleep onset latency rose by 14.2 minutes, WASO rose by 26.2 minutes and sleep efficiency fell by 9.5%. At 8 hours, WASO and sleep efficiency were still significantly worse. At 8 and 12 hours, total sleep time was about half an hour shorter on average, but those differences were not statistically significant. At 12 hours, latency to persistent sleep (time to the first 10 minutes of unbroken sleep) was 15.3 minutes longer. [1]

“A high dose of caffeine (400 mg) consumed as one dose in the morning is detrimental to subsequent night-time sleep, with greater disruptions the closer consumption occurs to bedtime.”

Gardiner et al. · Discussion · full text

N3 sleep dropped at every timing with 400 mg. Compared with placebo, N3 sleep was an estimated 20.6 minutes shorter when 400 mg was taken 12 hours before bed, 15.3 minutes shorter at 8 hours and 29.7 minutes shorter at 4 hours. The share of light sleep (N1 and N2) rose at 4 and 12 hours. [1]

“Therefore, the findings of the present study highlight the potent ability of caffeine to reduce N3 sleep.”

Gardiner et al. · Discussion · full text

What people felt didn't always match the recordings. Perceived sleep quality fell only when 400 mg was taken 4 hours before bed. Perceived total sleep time was shorter at 4 and 8 hours. Perceived WASO did not change significantly, even though measured WASO rose at 4 and 8 hours. [1]

“Importantly, consumers are unable to perceive sleep disturbance when consumption occurs further away from bedtime (i.e. 8 and 12 hours).”

Gardiner et al. · Practical Recommendations · full text

The authors' recommendation. Based on these results, they write that 100 mg can be consumed up to 4 hours before bedtime, and recommend avoiding 400 mg taken as one dose within 12 hours of bedtime. [1]

What this doesn't tell us

Everyone in the study was a healthy man aged 18 to 40 who slept well and drank a moderate amount of caffeine. The trial did not study women, older adults or people with insomnia. The authors note that older adults (over 40) have been shown to be more susceptible to caffeine-related sleep disruption, and that oral contraception is known to increase the half-life of caffeine. [1]

“Third, the study employed a healthy young adult male population which may limit the generalizability of the findings to alternative populations.”

Gardiner et al. · Limitations · full text

Each dose was taken all at once. The authors say it remains unclear whether the same total amount spread across the day would have the same effect. Only two doses and three times were tested. [1]

Sleep was recorded at home for one night per condition. Partners, children, pets, noise, light and diet were not controlled, and the prior night's sleep and time awake were not accounted for in the analysis. The authors advise caution when reading sleep-stage changes from a single night with the Somfit. [1]

The sample was small, and some 400 mg results did not reach statistical significance at every timing, including total sleep time at 8 and 12 hours. Participants' caffeine-related genes (CYP1A2 and ADORA2A) were tested, but the study was not powered to detect genetic effects. [1]

What to take away

Opinion, not a study result: If I were someone who sleeps poorly and drinks several strong coffees in one sitting, I would look at how much caffeine I take at once, not only at the time of the last cup. This trial tested healthy young men, so results may differ for others. If sleep problems persist, talk to a doctor.

Reference

  1. Sleep 2025[1] Gardiner CL, Weakley J, Burke LM, Fernandez F, Johnston RD, Leota J, Russell S, Munteanu G, Townshend A, Halson SL. Dose and timing effects of caffeine on subsequent sleep: a randomized clinical crossover trial. Sleep. 2025;48(4):zsae230.
    DOI: 10.1093/sleep/zsae230 · Full text

This article is for education only and is not medical advice. If sleep problems persist, talk to a doctor about your options.