Sleep Loss and Hormones: Testosterone, Insulin and Appetite

Most things that get blamed for “hormone problems” are hard to test. Short sleep is the exception. Because a night can be shortened in a lab with food and activity held constant, it has the tight before-and-after experiments in healthy people that diet and stress never get. The samples are small and the conditions artificial, but the results point one way.

Below: what those experiments found for testosterone, insulin, cortisol and the appetite hormones leptin and ghrelin, where the evidence is thinner than the headlines, and why the fix that has actually been tested is more sleep rather than a supplement.

Testosterone: eight nights of five-hour bedtimes

The study most quoted is a 2011 JAMA research letter. Ten healthy men with a mean age of 24 spent three nights with 10 hours in bed, then eight nights with 5 hours in bed, with blood sampled around the clock. After the restriction week, daytime testosterone was 10–15% lower, with the gap most obvious between 2 PM and 10 PM. Self-rated vigour fell in step, from 28 after the first short night to 19 after the seventh (Leproult 2011).

Ten men is a convenience sample, so two randomized studies from another group matter. Five inpatient nights of 4 hours in bed did not change testosterone against 9 hours in 14 men. A six-week outpatient trial in 13 men who cut habitual sleep by 1.5 hours a night found testosterone slightly lower overall but climbing back over the weeks, and the authors concluded that restriction did not adversely affect it (Smith 2019). A 2021 meta-analysis of 18 studies in 252 men drew the same line: short-term partial sleep restriction had no statistically significant pooled effect (standardised mean difference −0.22, confidence interval crossing zero), while total sleep deprivation of 24 hours or more lowered testosterone by a moderate amount (−0.64) (Su 2021).

The fair reading: a sustained run of roughly five-hour nights probably nudges daytime testosterone down modestly in young healthy men, one or two short nights probably do not, and nobody has shown the change reaching muscle, libido or mood beyond the vigour scores. Hold that number next to the supplement trials reviewed in Tongkat Ali and Testosterone before assuming a capsule is the right lever. All of this is in healthy men; anyone on testosterone therapy or with a diagnosed sleep disorder should ask their clinician rather than extrapolate.

Insulin: a single short night is enough

Insulin is where the sleep experiments are most consistent. Nine healthy adults (five men, four women) were each given a hyperinsulinaemic clamp twice: after a normal night of about 7.5 hours of sleep and again after a night of just under 4 hours. After the short night, glucose disposal fell by roughly a fifth (32.5 versus 40.7 µmol per kg of lean mass per minute), the liver released more glucose despite the insulin, and circulating fatty acids were higher (Donga 2010). One night.

A week gives a similar picture. Twenty healthy men aged 20 to 35 spent seven nights with 5 hours in bed after at least eight nights of 10-hour bedtimes, with diet and activity controlled in an inpatient unit. Insulin sensitivity measured by intravenous glucose tolerance test fell by about 20%, and by clamp by about 11%; glucose tolerance and the disposition index, a measure of how well insulin secretion compensates, also dropped. A modafinil-versus-placebo arm run during the restriction week made no difference (Buxton 2010).

The effect reaches into tissue. In a randomized crossover with seven young adults, four nights of 4.5 hours in bed versus 8.5, fat cells taken by biopsy needed nearly three times as much insulin to reach half of their maximal signalling response, and the overall response was 30% lower (Broussard 2012). A 2019 meta-analysis of 41 randomized sleep-restriction trials in healthy adults pooled the change in insulin sensitivity as a moderately large reduction (standardised mean difference −0.70) (Zhu 2019).

None of these participants had diabetes, and these are days-long experiments: they show that short sleep pushes a healthy metabolism in the insulin-resistant direction quickly, not what years of it do.

Cortisol: the evening problem

Cortisol is less dramatic than social media suggests. In young men, one night of partial sleep loss (sleep allowed only from 4 to 8 AM) or total sleep loss raised plasma cortisol the following evening by 37% and 45% respectively, and delayed the point at which cortisol quietens for the night by at least an hour. The change was detectable only in the evening (Leproult 1997). The week-long restriction study above recorded a roughly 50% rise in salivary cortisol too (Buxton 2010).

An evening rise is the wrong time of day: cortisol normally reaches its quiet period around bedtime before the morning cortisol rise, and the 1997 authors read the change as a slower wind-down from the morning peak rather than a new surge. A rise of roughly a third for a few evening hours in healthy young men is a delayed shutdown, not “adrenal burnout”; the authors’ own caveat was that repeated evening elevations might carry longer-term costs, which one night cannot show. It also probably matters less for insulin than intuition suggests: in the week-long study, changes in cortisol did not track changes in insulin sensitivity (Buxton 2010).

Appetite hormones: leptin down, ghrelin up, sometimes

The 2004 Annals of Internal Medicine study is the one usually behind the “sleep loss makes you hungry” story. Twelve healthy young men were studied after two days of restricted sleep and two days of extended sleep under controlled conditions. After the short nights, leptin was 18% lower, ghrelin 28% higher, hunger ratings 24% higher and appetite 23% higher, with the largest increase in appetite, 33–45%, for calorie-dense, high-carbohydrate foods (Spiegel 2004). Leptin signals fullness and ghrelin drives hunger; both moved the wrong way.

Outside the lab, among 1,024 adults in the Wisconsin Sleep Cohort, the models put leptin 15.5% lower with habitual 5-hour versus 8-hour sleep, and ghrelin 14.9% higher with 5 versus 8 hours of measured overnight sleep, and below 8 hours, shorter sleep went with higher body mass index (Taheri 2004). That is observational and cannot show cause.

The hormone story is shakier than the behaviour story. The 2019 meta-analysis of randomized trials found no strong evidence that sleep restriction changes average leptin or ghrelin, yet it did find that restriction raised subjective hunger by about 13 points on a 100 mm scale and intake by roughly 250 kcal a day, with no clear change in energy expenditure and a pooled weight gain of about 0.34 kg across the partial-restriction trials (Zhu 2019). So people reliably eat more after short sleep; whether leptin and ghrelin are the reason is unresolved.

Why it feels like a hormone problem

Put the pieces together and the feel of a bad week of sleep, flat, hungry, craving carbohydrate, less drive, wired in the evening, maps onto lower daytime testosterone, an insulin-resistant afternoon, a delayed cortisol shutdown and a bigger appetite. None of the individual changes is large in absolute terms: the testosterone gap in the JAMA letter was 16.5 against 18.4 nmol/L across the waking day. What the experiments show is that they arrive together, from a single input, within days.

It also explains why blood tests taken during poor sleep can mislead: a low-normal testosterone or a poorer glucose-tolerance result measured after a run of five-hour nights may partly be measuring the sleep rather than the person.

What reverses it

The intervention that has actually been tested is more sleep, and it works on the same timescale.

In a randomized crossover of 19 men who habitually slept a little over six hours on work nights, three weekend nights with 10 hours in bed, compared with 6, left insulin sensitivity higher and fasting insulin and leptin lower, and testosterone rose; morning cortisol did not change (Killick 2015). Three nights.

For appetite, the best evidence is a 2022 randomized trial in JAMA Internal Medicine. Eighty adults with overweight who habitually slept under 6.5 hours were assigned either to a single individualised sleep-hygiene counselling session aimed at 8.5 hours in bed, or to no change, at home. The extension group slept about 1.2 hours more a night and, with intake measured by doubly labelled water rather than food diaries, ate about 270 kcal a day less than controls (95% CI 147 to 393 kcal), lost weight relative to controls, and showed no difference in energy expenditure (Tasali 2022). No supplement, no diet, one conversation about bedtime.

That conversation is ordinary sleep hygiene, covered in the sleep hygiene basics post and, as a step-by-step sequence, in The Sleep Optimization Protocol.

Bottom line

Eight nights of five-hour bedtimes lowered daytime testosterone by 10–15% in ten young men; a single four-hour night cut insulin sensitivity by roughly a fifth in nine healthy adults; evening cortisol rises; hunger and intake go up, though leptin and ghrelin move less consistently than the popular story claims. The samples are small, but the direction is consistent and the reversal has been tested: three long nights improved insulin sensitivity in habitually short sleepers, and extending sleep by about 1.2 hours a night cut intake by about 270 kcal a day. If a hormone panel looks off after a bad month, more sleep before a repeat test is the cheapest experiment on offer. Anyone on hormone therapy, with a diagnosed sleep disorder, or managing a metabolic condition should discuss changes with a clinician.