![New Momo[Muscle] Logo](http://www.momomuscle.com/cdn/shop/files/momo-logo-73.png?v=1733197500&width=600){kind=logo}
By MoMo · Sourced + cited
Published August 24, 2026·Recovery·13 min read
Five nights of four-hour sleep cut the rate of new muscle protein by 19 percent — and training through those same five nights brought it back to normal. Here is what sleep loss actually costs, what it does not, and what to run on the days you did not get eight hours.
it is 1am and you are doing arithmetic. if you fall asleep now you get five hours and forty minutes. if you fall asleep in twenty minutes you get five hours and twenty. and somewhere in that math is the other question, the one that is actually keeping you up: is tomorrow's session even worth doing.
you have read that sleep is when muscle is built. you have read that growth hormone only releases in deep sleep. you have read a headline about testosterone dropping fifteen percent. so the conclusion assembles itself: bad sleep, wasted workout, might as well stay home and stop pretending.
that conclusion is wrong, and the studies that people quote at you are the ones that show it is wrong. i went and read them. some of the most-repeated claims in this topic trace back to a paper that literally uses the phrase "we hypothesized," and one of the most-quoted numbers comes from a sample of ten men.
so here is the whole thing. what sleep loss measurably costs, which mechanism does the damage, how long the growth window is actually open — that one is going to annoy you, because we had it wrong in our own topic brief — and the specific protocol for a session on four hours of sleep.
the straight answer up front. sleep loss costs you something real and it is smaller than you fear. five nights at four hours in bed lowered myofibrillar protein synthesis in young men from 1.53 to 1.24 percent per day — a 19 percent reduction, not a shutdown. and a third group who did the same five restricted nights while exercising ran at 1.61 percent per day, indistinguishable from men sleeping eight hours. the training defended the signal. so the answer to "should i skip it" is no. move it earlier, expect less from your heaviest compound, and go.
Key facts
| −19% | myofibrillar protein synthesis after 5 nights at 4 h in bed: 1.24 vs 1.53 %/day [Saner 2020, J Physiol] |
| 1.61 %/day | the same 4-hour nights plus exercise — back to control levels, and the single most useful number in this article [Saner 2020] |
| −7.56% | pooled performance decrement after sleep loss across 227 outcomes from 69 publications; morning tasks were largely unaffected [Craven 2022, Sports Med] |
| compounds only | consecutive restricted nights cut force output on multi-joint movements but not single-joint ones [Knowles 2018] |
| +34% | protein synthesis still above rest at 48 hours post-session in untrained lifters — the real ceiling of the "growth window" [Phillips 1997] |
| 60% more | fat-free mass lost when dieting on 5.5 vs 8.5 h in bed — 2.4 vs 1.5 kg over 14 days [Nedeltcheva 2010, Ann Intern Med] |
| 20–48% | lower all-cause mortality for the most regular sleepers vs the least, in 60,977 people — regularity beat duration [Windred 2024, Sleep] |
| 1.41× | the female-to-male risk ratio for insomnia across 1,265,015 people — this problem is not distributed evenly [Zhang & Wing 2006] |
what actually happens to muscle when you don't sleep
muscle mass is a balance between two ongoing processes: protein synthesis, where your body assembles new contractile protein, and protein breakdown, where it takes old protein apart. grow, and synthesis has outpaced breakdown across weeks. the specific number that matters for lifting is myofibrillar protein synthesis — the rate of building the actual contractile machinery, the myosin and actin that produce force, as opposed to the enzymes and structural proteins that make up the rest of a muscle cell.
until 2020, nobody had measured what sleep restriction does to that specific rate in humans. then a group at Victoria University in Melbourne did the experiment properly. twenty-four healthy young men, two nights of controlled baseline sleep, then five nights of intervention. they drank deuterium oxide — heavy water, which labels newly built protein so you can measure how fast it is being made — and gave muscle biopsies before and after.
three groups. normal sleep, eight hours in bed. sleep restriction, four hours in bed. and sleep restriction plus exercise, four hours in bed with three sessions of high-intensity intervals during the week.
the result, in fractional synthetic rate — the percentage of the myofibrillar protein pool rebuilt per day:
sleep restriction alone dropped the build rate by 19 percent. the same five four-hour nights with training in them ran at 1.61 %/day — not statistically different from men sleeping eight hours. the workout is not the thing sleep loss ruins. the workout is the thing that protects you from it.
sit with the shape of that for a second, because it inverts the advice you have been given. the standard version says sleep is when the building happens, so without sleep the training is wasted. the measurement says something closer to the opposite: without training, sleep loss quietly lowers your build rate — and training brings it back.
here is our own arithmetic on their numbers, and it is arithmetic rather than a finding, so treat it as such. the gap between normal sleep and restricted sleep is 0.29 percent of the myofibrillar pool per day. across the five nights of the study, that compounds to roughly 1.5 percent of your contractile protein that did not get rebuilt. one week of that is nothing you will ever see in a mirror. a year of it is a different conversation entirely, which is the honest way to hold this: sleep debt is a slow tax, not a switch.
the honest limits, stated plainly: twenty-four young men, four hours in bed is severe restriction rather than a normal bad week, and the exercise that rescued the rate was high-intensity cycling, not lifting. it is the closest direct evidence that exists on this question and it is still an inference to apply it to your squat session. the authors also looked for the molecular machinery they expected to explain the drop — the AKT and mTOR signalling everyone quotes — and found no changes in any of it. the rate moved. the mechanism people blame did not.
so should i skip the gym if i barely slept
no. but change three things about the session, and each change comes from a specific finding.
one: move it earlier. the largest analysis of this question pooled 227 performance outcomes from 69 publications and found a mean decrement of 7.56 percent after sleep loss, with a confidence interval running from 3.13 to 11.9 percent. but the subgroup analysis is where the usable instruction lives: tasks performed in the afternoon and evening were consistently impaired, while tasks performed in the morning were largely unaffected. the regression showed roughly a 0.4 percent decrease in performance for every additional hour spent awake before the task. by our arithmetic on that slope, a session at 8pm after being up since 5am is carrying about six percent more decay than the same session at 9am. the deficit is partly a function of how long you have been awake, and that is a variable you control.
two: keep the movements, retire the top set. a systematic review of seventeen studies on inadequate sleep and resistance exercise found something specific enough to program around: outright sleep deprivation had little effect on muscle strength, but consecutive nights of sleep restriction reduced force output on multi-joint movements — squats, presses, rows — and did not reduce it on single-joint movements. the more coordination and central drive a lift demands, the more sleep loss takes from it. your leg extension is fine. your back squat is the one that will feel like someone added plates while you were in the bathroom. so run the session, autoregulate by how the bar moves rather than by the number you wrote down last week, and take the heaviest single off the table. the pooled 7.56 percent decrement is the size of adjustment the evidence supports.
three: caffeine is a legitimate tool here, and it has a curfew. the same review names caffeine and training with other people as the two interventions that measurably restore performance under sleep loss. the curfew comes from a home-based trial that dosed 400 mg of caffeine at 0, 3, and 6 hours before bed and monitored sleep objectively. all three timings disrupted sleep significantly compared with placebo — including the one six hours out. so caffeine rescues today's session and, taken late, buys it by charging tonight's sleep. use it in the morning, not at 5pm, or you have simply moved the debt forward a day.
how long is the growth window actually open
this is where i have to correct our own brief. the working title for this article said "the 24 to 72 hour growth window." there is no 72-hour window. i went looking for the source of that number and could not find one, because the actual measurements say something tighter and more interesting.
the reference measurement comes from Phillips and colleagues in 1997, who infused labelled phenylalanine into eight untrained volunteers — four men, four women — and measured both synthesis and breakdown at rest and at 3, 24 and 48 hours after a single bout of resistance exercise. synthesis above resting levels: 112 percent at 3 hours, 65 percent at 24 hours, 34 percent at 48 hours. breakdown also rose, by 31 percent at 3 hours and 18 percent at 24, and was back to resting levels by 48. net balance — the number that decides whether you gain tissue — was improved at every timepoint out to 48 hours.
then MacDougall's group went looking for the far edge. they already knew synthesis was up 50 percent at 4 hours and 109 percent at 24. they measured six young men at 36 hours post-training, using the untrained arm as the control. at 36 hours the exercised arm was within 14 percent of the control arm, a difference that was not statistically significant. their conclusion, verbatim: synthesis rises rapidly, more than doubles at 24 hours, "and thereafter declines rapidly."
and then it gets one level deeper, which is the part almost nobody covers. Damas and colleagues in 2016 tracked ten young men through ten weeks of training and measured integrated synthesis at week one, week three and week ten. the biggest spike was week one — and it correlated with nothing. muscle fibres did not grow until week ten. what did correlate with growth, at r ≈ 0.9, were the smaller synthesis responses at weeks three and ten, once the muscle damage from being a beginner had faded.
the translation is unglamorous and useful: the enormous protein-synthesis spike a beginner gets is mostly repairing damage, not building new muscle. as you get trained, the window gets shorter and the spike gets smaller, and that is the response that actually turns into size. so when someone tells you a bad night's sleep wasted your 72-hour anabolic window, the correct response is that they have described a window that has not been measured, in a state that gets less relevant the better you get.
the claims worth keeping and the ones worth dropping
three of the most-quoted lines in this topic do not survive a read of their own sources.
What gets repeated
the sleep-and-gains canon, as it circulates
- "sleep debt breaks down muscle" — cited everywhere as established physiology
- "one week of short sleep drops testosterone 10 to 15 percent" — quoted to every audience
- "below 8 hours and your injury risk jumps 70 percent" — quoted as a general adult finding
- "no sleep, no gains" — the conclusion people actually act on
What the source says
read the papers, not the summaries
- that is a Medical Hypotheses paper whose own abstract says "we hypothesized" — a proposal, not a result
- ten men, mean age 24, 8 nights in bed 12:30–5:30am. real finding, narrow population, and for most readers here the least applicable line in this article
- 112 adolescent athletes, aged 12–18, odds ratio 1.7 with a confidence interval touching 1.0. a school-sports finding, not an adult one
- the one direct measurement showed a 19 percent slowdown that exercise reversed [Saner 2020]
the testosterone one deserves a sentence of its own, because it is quoted at women constantly by people who have not checked who was in the room. Leproult and Van Cauter put ten healthy young men on eight nights of five-hour bedtimes, measured total sleep at 4 hours 48 minutes, and found daytime testosterone fell to 16.5 from 18.4 nmol/L. they framed it against normal ageing, which costs 1 to 2 percent a year — so a week of short sleep did what a decade of getting older does. that is a genuinely striking finding in men. it is also the mechanism least likely to be doing the work in a body that does not run its muscle-building on those concentrations. the honest position is that we do not have the equivalent measurement in women, and pretending the male number transfers is how this topic became so full of confident nonsense.
which is the right moment to say the quiet part: almost every study in this article is men. Saner, twenty-four men. Leproult, ten men. the performance meta-analysis, 959 of its participants male — 89 percent. Nedeltcheva's diet study is the closest thing to balance here at three women out of ten. meanwhile the population most likely to be reading this is the one with a 1.41 times higher risk of insomnia across 1.27 million people. the group with the most sleep disruption has been studied the least on what it costs them in the gym. i cannot fix that with a blog post. i can refuse to hide it.
where sleep loss actually gets expensive: dieting
if there is one place to take this seriously, it is not the day after a bad night. it is a fat-loss phase.
Nedeltcheva and colleagues at the University of Chicago ran ten overweight adults through two 14-day blocks of the same moderate calorie deficit, in a crossover design, in a sleep laboratory. same diet both times. the only variable was 8.5 versus 5.5 hours in bed.
on 8.5 hours they lost 1.4 kg of fat and 1.5 kg of fat-free mass. on 5.5 hours, the same deficit produced 0.6 kg of fat and 2.4 kg of fat-free mass. same total weight lost. completely different composition. the short-sleep condition delivered 55 percent less fat loss and 60 percent more lean loss — and the participants reported being hungrier, with substrate use shifted away from burning fat.
that is the finding that should change a behaviour. a deficit is already a period where you are asking your body to give something up, and short sleep appears to change what it gives up. the caveats are real and worth stating: ten people, mean age 41, overweight and not lifting weights — no resistance training in the protocol at all, which matters enormously because lifting is the primary signal telling a body in a deficit to keep its muscle. two weeks is short. and "hours in bed" is opportunity, not measured sleep. it is one small elegant study, not a settled law. but the direction is the same direction everything else in this article points, and it is the only place where the numbers are large enough to reorganise a plan around.
the lever nobody sells you: regularity
every article on this topic ends by telling you to get eight hours, which is useless advice if the reason you are reading it at 1am is that you cannot. so here is the lever that is available to almost everyone, including shift workers, new parents, and people whose brains simply do not switch off.
researchers took more than 10 million hours of accelerometer data from 60,977 UK Biobank participants and computed a Sleep Regularity Index — a measure of how consistently sleep and wake happen at the same clock time, separate from how long they last. people in the four most regular quintiles had a 20 to 48 percent lower risk of all-cause mortality than the least regular quintile. and the headline finding: regularity predicted mortality more strongly than duration did.
be precise about what that does and does not show. it measured deaths, not muscle, in a cohort averaging 62.8 years old. it is observational, so it establishes association rather than cause. it is not evidence that regular sleep builds more muscle. what it is evidence for is that the variable everyone optimises — total hours — is not the only one that carries weight, and the other one is far easier to control. you cannot always decide to fall asleep. you can almost always decide when the alarm goes off.
and it happens to be the exact thing your training already runs on. your body adapts to what you repeat. unfortunately, that includes chaos.
the protocol
one thing to run tomorrow, built only from what is above.
| The lever | What to actually do | What it is worth, and from where |
|---|---|---|
| Train anyway | Do the session. Do not convert a bad night into a skipped week. | Exercise during 5 restricted nights held protein synthesis at 1.61 %/day vs 1.24 without it [Saner 2020, cycling intervals — inference to lifting] |
| Move it to the morning | Shift a bad-sleep session as early as the day allows. | Morning tasks largely unaffected; PM tasks consistently impaired; ~0.4% loss per hour awake [Craven 2022] |
| Autoregulate the compounds | Keep the exercises, drop the top set, judge by bar speed rather than last week's number. | Restriction cuts force on multi-joint but not single-joint lifts; pooled decrement 7.56% [Knowles 2018 · Craven 2022] |
| Caffeine, with a curfew | Use it pre-session. Nothing substantial inside 6 hours of your target bedtime. | Named as an effective countermeasure [Knowles 2018]; 400 mg at 6 h before bed still disrupted sleep [Drake 2013] |
| Anchor the wake time | Same alarm daily, including weekends. Let the bedtime float if it must. | 20–48% lower all-cause mortality for the most regular sleepers; regularity beat duration [Windred 2024 — mortality, not muscle] |
| Protect sleep hardest in a deficit | If you are dieting, this is the phase to defend 7+ hours rather than the phase to cut them. | 60% more fat-free mass lost on 5.5 vs 8.5 h at the same deficit [Nedeltcheva 2010, n=10, no lifting] |
| Eat before bed if it helps | Optional. 25–30 g of protein in the evening is reasonable, not magic. | 27.5 g pre-sleep protein over 12 weeks: +164 vs +130 kg strength in young men [Snijders 2015] — but no added benefit in older men [Holwerda 2018] |
and the sentence to keep, if you keep one: the bad night is already spent. the only decision still in front of you is whether you also skip the session, and that is the one the evidence is clearest about. a session on four hours of sleep is not a wasted session. it is a slightly worse session that protects the thing you are worried about losing. go early, go lighter on the big lift, and set the alarm for the same time tomorrow.
quick answers
is bad sleep killing my gains?
it costs you something measurable, but far less than the internet claims, and skipping the session makes it worse rather than better. in the only study to measure it directly, five nights with four hours in bed lowered the rate at which young men built new muscle protein from 1.53 to 1.24 percent per day, a reduction of about 19 percent. a third group did the same five restricted nights but performed three sessions of high-intensity interval exercise, and their rate was 1.61 percent per day, statistically indistinguishable from the men sleeping eight hours. exercise defended the muscle-building signal during sleep loss. one bad night is noise. a bad month is a real cost.
should i skip the gym if i barely slept?
no. train, but move it earlier and expect less from your heaviest compound lifts. a meta-analysis of 227 performance outcomes from 69 publications found a mean performance decrement of 7.56 percent after sleep loss, with roughly a 0.4 percent decrease for every additional hour spent awake beforehand. the same analysis found that tasks performed in the afternoon and evening were consistently impaired while tasks performed in the morning were largely unaffected. a separate systematic review of 17 studies found that consecutive nights of restricted sleep reduced force output on multi-joint movements like squats and presses, but not on single-joint movements. so keep the session, take the top set off the table, and do it early.
how long is the muscle growth window open after a workout?
shorter than most articles claim, and it shrinks as you get more trained. measured with stable isotope tracers in untrained people, mixed muscle protein synthesis was 112 percent above rest at 3 hours, 65 percent above at 24 hours, and 34 percent above at 48 hours after a single resistance session. a separate study measured it at 36 hours and found it had returned to within 14 percent of the untrained control arm, a difference that was not statistically significant. and in trained lifters the response is smaller: the synthesis spike measured in week one of a program was larger than the spike at week three or week ten, but only the later, smaller responses correlated with actual muscle growth. there is no 72-hour window.
does sleep timing matter more than the number of hours i get?
the timing appears to matter at least as much as the duration. using more than 10 million hours of accelerometer data from 60,977 UK Biobank participants, researchers found that people in the four most regular sleep-timing quintiles had a 20 to 48 percent lower risk of all-cause mortality than the least regular quintile, and that sleep regularity was a stronger predictor of mortality than sleep duration. that study measured deaths, not muscle, so it does not directly show a training effect. but it does mean that anchoring your wake time is a lever worth pulling on the nights you cannot control how long you sleep.
The reason a bad night turns into a skipped week is that there was never a fixed thing to show up to. Fighter is the fixed thing — the sessions are written, so on four hours of sleep you are only deciding to go, not deciding what to do.
GET THE PLANkeep reading
- when am i supposed to add weight. — the autoregulation this article asks for, written out properly
- why did i gain weight overnight. — the other thing a bad night does to a morning number
- how much protein do i actually need. — including where the pre-sleep protein question actually lands
- why don't i look like i lift yet. — the real timeline, which is longer than one bad week
- why can't i stay consistent. — for when the bad night became a bad month
References
| Source | What it supports |
|---|---|
| Saner et al. 2020, J Physiol 598(8):1523–36 — sleep restriction, HIIE and myofibrillar protein synthesis | n=24 young men, deuterium oxide, 5 nights: MyoPS 1.24 %/day restricted vs 1.53 normal sleep vs 1.61 restricted-plus-exercise. No change in AKT/mTOR or Foxo/LC3 markers — the rate moved, the assumed mechanism did not. Exercise was cycling intervals, not lifting — disclosed |
| Craven et al. 2022, Sports Med 52(11):2669–90 — acute sleep loss and physical performance | 227 outcomes, 69 publications: mean −7.56% [95% CI −11.9 to −3.13]; ~0.4% decrement per hour awake; PM tasks impaired, AM tasks largely unaffected. 959 of the participants [89%] were male — disclosed |
| Knowles et al. 2018, J Sci Med Sport 21(9):959–68 — inadequate sleep and muscle strength | 17 studies: consecutive restricted nights reduce force output on multi-joint but not single-joint movements; caffeine and group training named as countermeasures. Studies rated "moderate" or "weak" for global quality — disclosed |
| Phillips et al. 1997, Am J Physiol 273(1):E99–107 — synthesis and breakdown after resistance exercise | n=8 untrained [4 men, 4 women]: synthesis +112% at 3 h, +65% at 24 h, +34% at 48 h; breakdown +31% at 3 h, back to rest by 48 h; net balance improved at every timepoint to 48 h |
| MacDougall et al. 1995, Can J Appl Physiol 20(4):480–6 — time course of elevated synthesis | n=6 young men, contralateral control arm: synthesis +50% at 4 h, +109% at 24 h, back within 14% of control at 36 h [not significant]. The paper that closes the window |
| Damas et al. 2016, J Physiol 594(18):5209–22 — integrated MyoPS and hypertrophy | n=10, 10 weeks: the largest synthesis response was week 1 and correlated with nothing; only the smaller week-3 and week-10 responses correlated with hypertrophy [r ≈ 0.9], after muscle damage faded |
| Nedeltcheva et al. 2010, Ann Intern Med 153(7):435–41 — insufficient sleep and dietary fat loss | n=10 overweight adults [3 women], crossover, 14 days at the same deficit, 8.5 vs 5.5 h in bed: fat loss 1.4 vs 0.6 kg [−55%] and fat-free loss 1.5 vs 2.4 kg [+60%], plus increased hunger. No resistance training in the protocol — disclosed |
| Windred et al. 2024, Sleep 47(1):zsad253 — sleep regularity vs duration and mortality | 60,977 UK Biobank participants, >10 million hours of accelerometry: 20–48% lower all-cause mortality across the top four regularity quintiles, and regularity predicted mortality more strongly than duration. Mortality, not muscle; mean age 62.8; observational — disclosed |
| Leproult & Van Cauter 2011, JAMA 305(21):2173–4 — sleep restriction and testosterone | n=10 men, mean age 24.3, 8 nights in bed 12:30–5:30am [total sleep 4 h 48 min]: daytime testosterone 16.5 vs 18.4 nmol/L, a 10–15% fall against normal ageing of 1–2% per year. Men only — no equivalent measurement in women exists |
| Dattilo et al. 2011, Med Hypotheses 77(2):220–2 — sleep and muscle recovery | Cited across this topic as established physiology. It is a hypothesis paper: its own abstract states "we hypothesized that sleep debt decreases the activity of protein synthesis pathways." A proposal, not a finding |
| Drake et al. 2013, J Clin Sleep Med 9(11):1195–200 — caffeine at 0, 3 and 6 hours before bed | 400 mg of caffeine significantly disrupted sleep at all three timings, including 6 hours before bedtime, measured with a validated portable monitor. The empirical basis for the caffeine curfew |
| Zhang & Wing 2006, Sleep 29(1):85–93 — sex differences in insomnia | 31 studies, 1,265,015 participants: risk ratio for insomnia 1.41 [95% CI 1.28–1.55] female vs male, consistent and progressive across age |
| Milewski et al. 2014, J Pediatr Orthop 34(2):129–33 — sleep and sports injury | 112 athletes aged 12–18: sleeping <8 h was associated with 1.7× the odds of injury [95% CI 1.0–3.0]. Adolescents, survey-based, confidence interval touching 1.0 — quoted as adult fact elsewhere, which it is not |
| Mah et al. 2011, Sleep 34(7):943–50 — sleep extension in collegiate basketball players | Sleep extended by 110.9 min: sprint 16.2 → 15.5 s, free-throw +9%, three-point +9.2%. n=11 men, no control group — the direction is useful, the effect size is not transferable |
| Snijders et al. 2015, J Nutr 145(6):1178–84 — pre-sleep protein and training adaptation | 44 young men, 12 weeks, 27.5 g protein before sleep: strength +164 vs +130 kg, quadriceps CSA +8.4 vs +4.8 cm², type II fibre +2,319 vs +1,017 µm² |
| Holwerda et al. 2018, J Nutr 148(11):1723–32 — the same strategy in older men | 41 men aged ~70, 12 weeks: no further increase in muscle mass or strength from post-exercise and pre-sleep protein. The honest other half of the pre-sleep protein story |
| Watson et al. 2015, Sleep 38(6):843–4 — AASM and SRS joint consensus | The professional-body recommendation of 7 or more hours per night for adults, used here as the reference target rather than as a training claim |
| Alzueta & Baker 2023, Sleep Med Clin 18(4):399–413 — the menstrual cycle and sleep | Across regular, asymptomatic cycles objective sleep continuity does not change, though sleep EEG does and reported sleep quality is worse premenstrually and during menstruation in those with symptoms or cramps |
educational content, not medical advice. persistent insomnia, loud snoring with daytime sleepiness, or unrefreshing sleep despite adequate time in bed are clinician questions rather than training questions — untreated sleep apnoea in particular will not be solved by an earlier alarm. populations are named in the table above deliberately: the majority of the evidence in this article was collected in young men, the regularity data measured mortality rather than muscle, the injury data came from adolescents, and no direct measurement of sleep restriction and myofibrillar protein synthesis in women currently exists.
- MoMo