Sleep efficiency, and what the number actually means.
The share of your time in bed that you spend asleep. It is the number sleep clinicians work from, it takes two minutes a morning to record — and most people working it out at home get one of the four inputs wrong.
Sleep efficiency is total sleep time divided by time in bed, multiplied by 100. A figure of 85% or above is the threshold usually taken as normal sleep continuity; below it is generally read as sleep-continuity disturbance. It is the arithmetic underneath sleep restriction therapy, and it is deliberately crude — it counts minutes, and nothing else.
The four numbers, and the ones people get wrong
| What to record | What it means, and the mistake to avoid |
|---|---|
| Time in bed | From when you got into bed intending to sleep to when you finally got out of it. Reading, television and phone time before lights-out do not count — this is the commonest error, and it makes efficiency look far worse than it is. The lie-in after you woke does count, because you were still in bed. |
| Time to fall asleep | Your best estimate, made in the morning. Sleep researchers call this sleep-onset latency. |
| Time awake in the night | Everything you were awake for after first falling asleep, added together. Again, estimated in the morning. |
| Total sleep time | Time in bed, minus the two figures above. You never measure this directly — it falls out of the others. |
Estimate all of it in the morning, and do not check a clock during the night. That is not a rounding convenience. Clock-watching prolongs wakefulness, and CBT-I guidance advises against it explicitly. A rough figure written down at breakfast is the intended input; a precise one taken at 3am has cost you more than it measured.
Work out last night's figure
Nothing is sent anywhere — this runs in your browser and is forgotten when you close the page. It is here because the definitions above are the hard part, not the division.
What the number means
| Figure | How it is usually read |
|---|---|
| Under 85% | Below the usual threshold for normal sleep continuity. On one night this means very little. Sustained across two weeks, it is the pattern worth taking to a GP or a sleep service. |
| 85 – 90% | The normal range. |
| 90 – 97% | High. This is the range sleep restriction therapy aims at during active treatment. Outside treatment, read it alongside total sleep time. |
| Above 97% | Unusually high, and not a gold star. Healthy sleepers have brief awakenings they do not remember. Consistently near-perfect efficiency more often means too little time in bed than exceptional sleep. |
What it does not tell you
It is purely quantitative. It counts minutes asleep against minutes in bed, and says nothing about sleep architecture, how much deep or REM sleep you got, or how rested you feel. Efficiency and feeling rested genuinely do come apart, and when they do, the feeling is not the thing that is wrong.
Self-report drifts in a known direction. Measured against EEG, people tend to overestimate how long they took to fall asleep and underestimate how much they slept. That is expected, it is why a diary is still clinically useful, and it is another reason a single night is not worth interpreting.
Tracking it too closely can make sleep worse. There is a recognised pattern — orthosomnia — in which the pursuit of a perfect sleep score becomes its own source of arousal at bedtime and worsens the insomnia it was meant to fix. If recording the number is making you anxious about sleeping, that is a reason to stop recording it, not a reason to try harder.
Common questions
Is 100% sleep efficiency good?
No. Brief awakenings are normal and mostly unremembered. A consistent 100% usually indicates too little time in bed for the sleep you need.
Does time in bed include reading or watching TV?
No — only the period you were in bed intending to sleep. Counting the hour you spent reading is the commonest error, and it drags the figure down artificially.
Should I fill it in during the night?
No. In the morning, from memory. Checking a clock in the night prolongs wakefulness.
How many nights do I need?
Seven to fourteen. A baseline of a week or two is the standard before drawing any conclusion or changing anything, because night-to-night variation is large enough to swamp a single reading.
Where this comes from
Sleep efficiency is a foundational polysomnography measure and the mathematical basis of sleep restriction therapy, developed by Arthur Spielman in the 1980s. The 85% threshold, and the higher targets used during treatment, are standard in the American Academy of Sleep Medicine's clinical practice guidance on insomnia. The modern reference form for a prospective sleep diary is the Consensus Sleep Diary (Carney et al., 2012), developed with the AASM.
We describe the method and name its sources. We do not reproduce any clinical instrument or scale. Nothing on this page is medical advice or a diagnosis, and it cannot distinguish insomnia from the many other things that disturb sleep. If your sleep is persistently poor, or you are sleepy in a way that affects driving or work, see a GP.
Doing it for thirty nights
Everything above is free, and for a night or two a phone and the back of an envelope will do the job. The part almost nobody manages is doing it every morning for long enough that the number means something — which is the only condition under which it is worth calculating at all.
That is what The Sleep Journal is: thirty morning pages with the four boxes already drawn, the arithmetic laid out, a weekly checkpoint that adds it up, and a verdict at the end compared against the page you signed on night one. It carries the warnings on this page inside it, including the one about not driving on a restricted sleep window.