Sleep and the Trail Runner: How to Plan the Night Before It Plans You

Look at when your race starts. Not the date, the time. If the answer is anything later than about 14:00, you have signed up to run through the part of the twenty-four hour cycle when a human body is least willing to do anything, and you will probably do it having been awake since seven that morning, having never once rehearsed it.

Most training advice treats this as a gear problem. Buy a brighter headtorch, pack a spare battery, done. It is not a gear problem. It is a scheduling problem with a physiological cost, and unlike heat or altitude, it is one you can rehearse for free.

Ten of seventeen races start after lunch

Of the seventeen races on this site with a published start time for the next edition, ten begin at 13:00 or later. Six begin after 18:00. UTMB goes at 17:45. TDS at 23:50. SainteLyon at 23:30. Evening and night starts are not an exotic subgenre of trail running; in the mountain ultra calendar they are close to the default.

TrailMath already prices this. The race conditions simulator works out how much of your race falls between 21:00 and 05:00 and charges 20% of it. Feed it a few real start times and something strange comes out:

Race and finish time	Night fraction	Our grade	Nadirs crossed	Hours awake at the finish
SainteLyon, 23:30 start, 9h	66.7%	brutal	1	25.5
A 16:00 start, 12h	58.3%	brutal	1	21.0
TDS, 23:50 start, 30h	46.7%	brutal	2	46.8
UTMB, 17:45 start, 30h	33.3%	challenging	1	40.8
UTMB, 17:45 start, 46h	34.8%	challenging	2	56.8
Western States, 05:00 start, 29h	27.6%	challenging	1	30.5

Night fractions are what our model computes; it works in whole hours, so 17:45 goes in as 17. Nadirs are crossings of roughly 04:00. The last column assumes you woke at 07:00 and never napped, which for the three late starts describes a runner making the exact mistake this article is about.

Read the first and last rows together. A twelve-hour race starting at 16:00 is a more nocturnal race, by fraction, than a thirty-hour UTMB. Our own grader calls it brutal and calls a forty-six hour UTMB merely challenging. On darkness, that is correct. On sleep, it is backwards: the UTMB runner is awake for fifty-seven hours and crosses two circadian lows, and the twelve-hour runner is home and in bed before most people have finished breakfast.

Two different variables have been wearing the same name. Time in the dark is a visibility and footing problem. Sleep deprivation is a function of how long you have been awake and how many times you have crossed the bottom of your body clock. They diverge, hard, and every practical decision below follows from telling them apart.

What sleep does in a training block, stated honestly

The strongest evidence available is a 2023 meta-analysis by Lopes and colleagues: thirty-one studies, 478 participants, pooled effect of sleep deprivation on endurance performance of -0.52. Moderate, consistent, and with one finding that matters more to us than the headline number - efforts longer than thirty minutes were hit harder than shorter ones. Whatever this effect is, it scales with duration, which is bad news for a sport measured in tens of hours.

Martin (1981) is old and small, eight subjects, but it is the one that describes what you will actually feel. After thirty-six hours awake, treadmill time to exhaustion fell about 11%, and subjects rated the same workload as harder than they had rated it rested. The pace is not lying to you. Your sense of what the pace costs is.

The same study found something more useful than its average. Four subjects barely changed. Four fell apart, losing 15 to 40%. If you have ever wondered why one friend seems to shrug off a night out and another is a passenger for three days, that is the finding. There is no test for which one you are except doing it.

Sleep loss does not slow you down so much as it makes the same speed feel unaffordable.

Beyond that, be careful what you believe. The famous Stanford study where basketball players got faster after sleeping more was eleven athletes with no control group. The finding that under-eight-hours athletes get injured more often was a survey of 112 teenagers, with a confidence interval that touches 1.0. Growth hormone really is secreted mostly in deep sleep, but nobody has shown that losing some of it measurably blunts endurance adaptation. These are worth knowing and not worth leaning on.

One practical consequence: TrailMath’s plan generator does not know about any of this. It watches your RPE, your reported pain, and whether you are completing sessions. It has no field for sleep, no concept of time of day, and it cannot prescribe a night run. Everything in the rest of this article is something you add yourself.

Banking: the one intervention with a dose

Most pre-race sleep advice is vague encouragement. Sleep extension is the exception, because it has been tested properly and it comes with numbers.

Rupp and colleagues (2009) put one group at ten hours in bed for a week while a control group kept its habitual seven, then restricted both to three hours a night. The extended group degraded less and recovered faster. Arnal and colleagues (2016) ran the version closest to our sport: six nights at 9.8 hours in bed versus 8.2, then thirty-four to thirty-seven hours of continuous wakefulness. That is a UTMB. Sustained contraction endurance improved about 8% after the deprivation, and perceived exertion dropped about 7%. Roberts and colleagues (2019) got a roughly 3% time-trial improvement in trained endurance athletes from just three nights of extended sleep.

So the prescription is real: an extra one to two hours in bed for the last five to seven nights before the race. Two caveats. It is time in bed, not sleep achieved, and the studies had supervision that you will not. And it works on a dose, so a single virtuous early night in race week is not the intervention.

Runners already do this by instinct, badly. In a survey of 636 ultrarunners, 54.7% said they tried to extend sleep before a race, but only about a fifth had a structured plan for sleep during it. Poussel and colleagues, looking at UTMB 2013, found that runners who increased their sleep beforehand finished faster.

Here is the part nobody mentions. If your race starts in the evening, bank by getting up later, not by going to bed earlier. An earlier bedtime shifts you the wrong way, toward being a morning person, three days before an event that will ask you to be awake at four. Extending the back end of your night gives you the same hours in bed and nudges your body clock in the direction the race needs.

Race day when the gun goes at 17:45

First, what not to do. You will find advice about shifting your body clock in the week before a night race. The protocols that genuinely achieve this come out of shift-work research and involve intermittent bright light around 5000 lux for hours, dark glasses on the way home, sustained-release melatonin, a blacked-out bedroom, and five consecutive nights of discipline. That is a clinical intervention. A runner who does a casual half-version of it arrives at the start line circadian-misaligned and short of sleep, which is worse than doing nothing.

What works is duller and more reliable.

Sleep in. You are unlikely to sleep well the night before anyway, and that is fine; the evidence that matters sits in the whole week, not the last night. In one study of a 217 km race, finishers actually took longer to fall asleep the night before than non-finishers did. Do not read anything into a bad night.

Take a real nap in the early afternoon. For a 17:45 start this is the single highest-value thing you will do all day, and almost nobody does it, because the race village is exciting and the pasta is free. Ninety minutes, curtains shut, alarm set.

Treat the last few hours as a shift handover. You are about to start a night shift, and you have spent the day travelling, queueing for kit check, and being nervous.

Then there is the fuelling wrinkle, which is specific to evening starts and which I have never seen written down. Your last unhurried meal is lunch, roughly five hours before the gun. Everything after that is nervous grazing. More importantly, the night appetite crash that UTMB runners hit around Courmayeur, twelve to fifteen hours in, arrives at hour five or six of a twelve-hour evening race - while you are still running hard, not while you are already hiking. Plan the switch to savoury food, broth and salt for the point where the clock says 22:00, not for some fraction of the way through.

The same clock problem applies to caffeine, and this one is worth being precise about.

03:00, and what to do about it

Core body temperature bottoms out around three in the morning, and alertness, reaction time and mood go with it. This is the part of the night people mean when they talk about the wall.

Hurdiel and colleagues (2018) measured it during a 168 km mountain ultra: ninety-two runners, sleepiness scales and a cognitive task at eight checkpoints. Decrements got worse as the race went on, and were modulated by time of day, with the worst readings in the early morning hours. Not a lab. That race.

What runners actually do about it is mostly nothing. Hurdiel’s earlier work followed seventeen UTMB finishers over races of 27 to 44 hours and recorded a mean total rest of twelve minutes, with a standard deviation of seventeen. And in the largest dataset we have, Kishi, Millet and colleagues surveyed 1,154 finishers of the Diagonale des Fous and the Trail de Bourbon: 77% slept at some point, cumulative sleep averaged 76 minutes on the 165 km race and 27 minutes on the 111 km one, 58% used some kind of strategy, and 80% reported at least one symptom they attributed to sleep deprivation, including falls and hallucinations.

On naps, the evidence has just improved. Bourlois and colleagues (2026) studied 52 UTMB finishers across the 2022 to 2024 editions, offering a twenty-minute sleep opportunity at three mid-race checkpoints. Perceived sleepiness dropped almost immediately; the cognitive benefit took longer to appear. Crucially, napping did not compromise finish times. Twenty minutes is not free, but it is close enough to free that the argument against it is weak.

Three rules follow.

Twenty minutes, not more. Past about half an hour you drop into deep sleep, and waking out of it buys you grogginess that can last longer than the nap.

Decide the trigger points before you start. Runners choose whether to nap based on how sleepy they feel, and feeling sleepy is a poor proxy for being impaired - Bourlois found subjective alertness recovers before objective vigilance does. You will feel sharp before you are sharp, which matters if the next section is a technical descent. Write the aid stations where you are allowed to nap onto your aid station plan in advance, and let the plan decide, not the version of you that has been awake for a day and a half.

The reason most ultra naps fail is temperature. You stop, you are damp, it is the coldest hour, and you lie there shivering instead of sleeping. Dry layer first, then something hot, then lie down. There is no trial behind this one; it is a practical heuristic that experienced runners converge on.

How much to plan depends entirely on distance, and the 636-runner survey gives clean brackets. Under 36 hours, most runners sleep not at all and probably should not plan to. Between 36 and 60 hours, about half sleep, typically well under two hours. Beyond 60 hours - Tor des Géants territory - 95% sleep, averaging over eight hours, and the question stops being whether and becomes where. Plan those at the life bases.

One honest caveat so the naps are not oversold: runners who had napped before testing scored worse on cognitive tasks, most likely because people nap when they are already wrecked. A nap is a safety and alertness intervention. It is not a performance enhancer.

The debt comes due after the finish

You will feel roughly human within about three days. That feeling is not reliable. Work on elite adventure racers by Mann and colleagues (2024) found that objective sleep debt had not returned to pre-race levels within a week, while subjective fatigue and cognition normalised much sooner. The second night after a big race is often more broken than the first, because now everything hurts.

TrailMath’s post-race Recovery block is calibrated on musculoskeletal repair. It has no sleep term in it at all, which is worth knowing when it tells you that you are ready.

There is a last tension worth naming, because it cuts against advice elsewhere on this site. In our piece on heat and altitude we tell you to arrive early and sleep high before a mountain race, and for aerobic adaptation that is right. But sleeping high degrades sleep: at 2,590 m, forty-four healthy men went from a median apnoea-hypopnoea index of 4.6 per hour at low altitude to 13.1 on their first night, with slow-wave activity down about 15%. The first night is the worst and it improves, so arriving several days early is better than arriving two. But you cannot fully bank sleep and fully acclimatise at the same time, and we would rather tell you that than pretend the two plans stack neatly.

Which is the whole point. Sleep is not the variable you tough out at three in the morning. It is one you plan, weeks earlier, with the same seriousness you give to vertical gain and carbohydrate per hour - and then, when the low point arrives on schedule and your legs feel fine but the trail has started rearranging itself, you find out whether the planning worked.