Should One Bad Night's Sleep Cancel Your Interval Session?

You wake up at 6:10 a.m. after four hours and forty minutes of sleep. The baby, the deadline, the neighbour’s party, whatever it was. Your Garmin says Body Battery peaked at 34. Runna has 6 x 800m at 5K pace scheduled for tonight. Do you bin it?

Short answer: no. Not on the strength of one night. But it’s worth being precise about why, because the same logic tells you to do something quite different when the bad nights stack up.

So, does sleep affect running performance?

Yes, but the size of the effect depends heavily on what kind of effort you’re asking for and how long the deficit lasts.

The research is fairly consistent on the shape of it. A single night of restricted sleep (say 4 to 5 hours) tends to raise perceived effort at a given pace. Time-trial and maximal-effort performance can dip by a few percent in some studies, and in others barely moves. Short, sharp, well-rehearsed efforts are the most robust. Long, grinding efforts where you have to tolerate discomfort for an hour or more are the most fragile, mostly because your head gets in the way before your legs do.

Chronic restriction is a different animal. Weeks of sleeping 5 to 6 hours when you need 7 to 8 are tied to slower reaction times, worse mood, higher resting heart rate, suppressed HRV and a higher injury and illness risk. That’s no longer a bad day. That’s a state, and your training is happening inside it.

Hold on to that distinction. It does most of the work in this post.

One night: adjust the expectation, keep the session

Here is the practical version. After a single short night, the session is usually still worth doing. What changes is the scoreboard.

Take a runner whose Runna plan says 6 x 800m at 4:20/km with 90 seconds jog recovery. Normally their rep splits land between 3:28 and 3:32. After the four-hour-forty night:

Notice that the session still happens. You’re still giving your body the stimulus: legs turning over quickly, aerobic system working near VO2max. A few seconds a rep slower does not turn that into a wasted workout. The adaptation comes from the intensity relative to your state, not from hitting a number on a screen.

There’s also a quieter benefit. If you cancel every time a night goes wrong, you teach yourself that sleep is a veto, and for a parent or a shift worker that means you never train. Consistency beats perfection by a large margin.

The Garmin and Whoop trap

Wearables can nudge you toward cancelling when you shouldn’t. Garmin’s Training Readiness might show 41 (“Low”) after one rough night even though your HRV status is still “Balanced” and your acute load is modest. That score is a blend: sleep score, recovery time, HRV, sleep history, stress. A single input can drag the headline number down while the underlying picture is fine.

Before you act on a low number, check three things:

1. Overnight HRV against your 7-day baseline. One low reading is noise. Garmin’s balanced range is built on weeks of data for a reason.
2. Resting heart rate. If it’s within 3 bpm of normal, you are probably fine. If it’s up 7 or 8 bpm, something more than sleep is going on (illness incoming, alcohol, a hard week).
3. How the first ten minutes feel. Honestly the best sensor you own. Legs heavy and heart rate drifting up on an easy jog is information. Feeling flat but stable is just tiredness.

I go deeper on how to weigh these in Reading Readiness, HRV and Recovery Scores, but the rule of thumb is: one metric is a prompt to look, not a verdict.

When to genuinely cancel or swap

There are cases where a single bad night should change the plan, not just the expectations:

• Under about 3 hours of sleep, or a night awake entirely. Swap the intervals for 30 to 40 minutes easy, or rest.
• You feel ill. Sore throat, fever, a chesty feeling. Sleep loss plus illness is where people get hurt.
• The session is a race-pace long effort (for example 3 x 3km at half-marathon pace in marathon week). Those are brutal when tired, and a day’s delay costs you almost nothing.

Moving a session by 24 hours is not cancelling it. Swapping Thursday’s intervals and Friday’s easy run is a perfectly legitimate edit, and most apps, including Runna and TrainingPeaks, let you do it in two taps.

A run of poor weeks: restructure the block

Now the other scenario, and the one that I think runners underreact to.

Picture a runner on a 16-week ChatGPT-built half marathon plan. Weeks 6 to 9 coincide with a work crunch. Their Strava log and Garmin sleep data look like this:

Week  Avg sleep  Avg RHR  HRV (7d avg)  Easy pace @ HR 142   Quality sessions hit
 5    7h 05m     51       62 ms         5:52/km              2 of 2
 6    6h 10m     53       58 ms         5:55/km              2 of 2
 7    5h 40m     55       52 ms         6:03/km              1 of 2
 8    5h 50m     56       49 ms         6:09/km              1 of 2
 9    5h 35m     57       47 ms         6:14/km              0 of 2

No single night in that table is the problem. The trend is. Resting heart rate is up 6 bpm, HRV has fallen about 24 percent, and the same easy heart rate now buys 22 seconds per km less speed. The runner is missing sessions and, worse, grinding through the ones they do finish.

The plan, meanwhile, will keep marching on. Week 10 says 48 km with a tempo and long run with a fast finish. The AI doesn’t know you’ve slept five and a half hours for a month. It only knows the schedule.

If the same plan were handed to a coach, they would not say “try harder.” They would restructure:

1. Repeat a lower-load week. Drop back to roughly week 5 volume (about 70 to 75 percent of week 9) for 7 to 10 days.
2. Keep one quality session, not two. One moderate tempo or short intervals per week holds fitness without adding much fatigue.
3. Protect the long run’s length, not its intensity. Cut the fast finish.
4. Re-enter the plan one step back, not where the calendar says you should be. Losing a fortnight of progression costs far less than a stress fracture or a flu in week 12.
5. Move the goal if needed. If your race is in six weeks and you have lost three weeks of good training, a realistic target is last year’s time, not the PB.

Then fix the actual input. Aim for a consistent wake time, even at weekends. Treat 7 hours as a training requirement, not a luxury. If the next four weeks of data (sleep, RHR, HRV) trend back toward baseline, resume progression.

A simple rule you can paste into any plan

Whether your plan comes from Runna, Garmin Coach or a ChatGPT prompt, you can give yourself this decision rule:

• 1 night under 6 hours: run the session, loosen the targets by 3 to 5 percent, warm up longer.
• 2 to 3 nights under 6 hours in a row: keep the easy runs, convert the hardest session to a steady run or move it a day or two.
• Under 6.5 hours average for 7+ days, with RHR up 4+ bpm or HRV down 10+ percent: drop the load, simplify the week.
• Two or more weeks like that: the block is no longer fit for you. Restructure it.

If you use ChatGPT to build plans, feed it the evidence instead of a feeling. Something like: “Last four weeks: average sleep 5h 45m, resting HR up from 51 to 57, HRV 7-day average down from 62 to 47 ms, easy pace at 142 bpm slower by 22 s/km. I have a half marathon in 7 weeks. Rebuild weeks 10 to 16 with a down week first.” A model with those numbers will give you a far saner plan than one that only knows your goal time.

Tonight, then: you slept four hours forty. Warm up for fifteen minutes, run the first rep by feel, and let the watch be a witness rather than a judge. And if you’ve been sleeping like this since September, the interval session isn’t the question. The plan is.