How to Keep a Sleep Diary (and Why It Is the First Step)

6:42 am. I slept terribly. Or at least, I think I slept terribly. The truth is that I am not entirely sure how much I slept, how long I was awake, or whether that stretch from 1am to 3am was two hours of wakefulness or thirty minutes that felt like two hours. My memory of last night is a montage of clock faces and ceiling textures, and it is probably wrong.

This is the first reason to keep a sleep diary: your perception of your own sleep is unreliable.

I resisted the diary for the same reason I resisted most of CBT-I, which is that it seemed too simple to address something so entrenched. Writing down what time I went to bed and what time I got up felt like bringing a clipboard to a house fire. The fire, however, turns out to have a pattern, and the clipboard is how you find it.

Why memory lies about sleep

Harvey and Tang’s research on sleep perception shows that people with insomnia systematically overestimate how long they lay awake and underestimate how much they slept [3]. The discrepancy is not small. Someone who reports sleeping four hours may have slept six, measured objectively. The wakeful minutes are vivid and memorable; the sleeping minutes, by definition, are not.

This perceptual distortion operates through several pathways. At 2am, time dilates. A twenty-minute period of wakefulness, spent in the dark monitoring your own failure to sleep, feels like an hour. The attention is selectively tuned to wakefulness: you remember the ceiling, the clock, the thoughts, and you do not remember the stretches of light sleep in between because light sleep does not produce memorable experiences. The result is a narrative in which the night was a continuous ordeal, when objectively it was an ordeal with intervals of actual sleep that your memory discarded.

This matters for treatment because CBT-I, specifically the sleep restriction component, requires an accurate baseline. If you think you are sleeping four hours and restrict to a four-hour window, but you are actually sleeping six, you are restricting too aggressively. The sleep diary, filled out in the morning rather than recalled later, provides a closer approximation of reality than memory alone.

It also matters psychologically. The insomnia narrative (“I never sleep,” “every night is terrible”) feeds catastrophic thinking, which feeds arousal, which feeds insomnia. The diary replaces the narrative with data, and the data almost always shows that the reality, while bad, is less uniformly catastrophic than the story.

What to track

The Consensus Sleep Diary, developed by Carney et al. in 2012, standardized what had been a variety of informal formats [1]. The core fields are:

Time into bed. When you actually got into bed, which may be different from when you decided to sleep.

Time tried to sleep. When you turned off the light and intended to fall asleep.

Sleep onset latency. Your estimate of how long it took to fall asleep. This will be imprecise. That is acceptable.

Number of awakenings. How many times you woke during the night.

Total wake time during the night. Your estimate of total minutes awake across all nighttime awakenings.

Final wake time. When you woke for the last time.

Time out of bed. When you actually got up.

Sleep quality. A subjective 1-5 rating.

Optional fields: caffeine intake (amount and time), alcohol, naps (time and duration), sleep medications.

The list looks straightforward. Two things trip people up. First, “time into bed” and “time tried to sleep” are different entries because many people get into bed and scroll their phone, read, or lie there for thirty minutes before attempting sleep. That gap between getting into bed and trying to sleep is significant; it reveals how much of your “time in bed” was never time trying to sleep at all. Second, sleep onset latency is an estimate. You are not supposed to know exactly when you fell asleep, because knowing exactly would require monitoring, and monitoring keeps you awake. Rough is fine. The diary works on averages.

When and how to fill it in

Fill in the diary within thirty minutes of getting up in the morning. Not during the night. Checking the time during the night to record it defeats the purpose; clock-monitoring is one of the behaviors that maintains insomnia. Estimates are fine. The diary is looking for patterns over two weeks, not minute-by-minute precision on any single night.

This instruction is harder to follow than it sounds. The temptation is strong: you wake at 2:17am, check the clock, and think “I should note that for the diary.” That check is not neutral. It triggers a time calculation (“four hours until the alarm”), which triggers an assessment (“that is not enough”), which triggers arousal (“tomorrow is going to be terrible”), which ensures you stay awake longer. The diary entry is not worth the cognitive cost of the clock check. Estimate in the morning. The estimate will be wrong by ten or twenty minutes. Over two weeks of estimates, the average will be close enough.

Paper or app: either works. The CBT-I Coach app, developed by the US Department of Veterans Affairs, has a built-in diary format and is free. A paper form on the nightstand works equally well and has the advantage of keeping screens away from the bed. I used paper because the phone was already banned from the bedroom and I did not want to create an exception for it.

What the numbers tell you

After two weeks, calculate your averages:

Average total sleep time (TST). Add up your estimated sleep per night and divide by the number of nights.

Average time in bed (TIB). Bed to rising, averaged.

Sleep efficiency. TST divided by TIB, expressed as a percentage. If you average six hours of sleep in eight and a half hours in bed, your sleep efficiency is about 71%.

Sleep efficiency below 85% indicates significant time spent awake in bed [4]. This number becomes the basis for your sleep restriction window. Your initial prescribed time in bed will equal your average total sleep time (with a floor of five hours), designed to consolidate your sleep into a more efficient, less fragmented block.

The floor matters. No matter how low your diary average comes out, the starting window does not go below five hours. This is a safety measure: below five hours, the daytime impairment becomes too significant and the treatment becomes counterproductive. If your diary shows four hours of average sleep, you still start at five.

What the diary reveals that you didn’t expect

Most people who start a sleep diary discover one or two things they were not expecting. Common revelations:

You are sleeping more than you thought. The gap between perception and diary data is often large enough to reframe the problem. You may still have insomnia, but the severity may be different from what 3am told you. I discovered I was averaging about five and a half hours, not the three and a half I would have guessed. Still not enough, but a different kind of not enough.

Your time in bed is much longer than your sleep time. Nine hours in bed with six hours of sleep is a common pattern, and it is the pattern sleep restriction targets. The three extra hours are not bonus recovery time. They are hours spent practicing wakefulness in the place where you are trying to sleep.

There are nights that are actually fine. Insomnia tends to colonize your narrative: every night is bad, nothing works, you never sleep. The diary shows that some nights are worse and some are better, which is both more accurate and more treatable. A problem with variability is different from a problem with total absence, and the treatment approaches reflect that.

Your weekend and weekday patterns are different. Many people sleep later on weekends, creating the social jet lag that shifts the circadian clock and worsens Monday and Tuesday nights. The diary makes this visible in a way that memory, which blends nights together, does not.

What makes this hard

The diary is the easy part of CBT-I, but it still has friction. Estimating sleep onset latency feels like guessing. Estimating total wake time feels like guessing. The temptation to check the clock “just for the diary” is real and counterproductive.

Accept that the estimates are rough. Two weeks of rough estimates produce a useful average, even if any single night’s entry is off by twenty minutes. The diary is a population-level tool applied to your population of nights. A single night’s data tells you almost nothing. Fourteen nights, even imprecisely recorded, tell you something useful.

There is also an emotional friction. The diary confronts you with data about a problem you would rather not look at. Writing “time to fall asleep: 90 minutes” every morning is not pleasant. Neither is seeing a sleep efficiency of 62% staring back at you after two weeks. The data does not feel like progress. It feels like documentation of a problem you already knew was bad.

The reframe that helped me: the diary is not a report card. It is a baseline measurement. A doctor measuring your blood pressure before prescribing treatment is not judging you for the number; they are establishing where you start so they can track where you go. The diary does the same thing for sleep. The numbers are not a verdict. They are coordinates.

The harder part is what comes after. The diary gives you numbers. The numbers tell you what your sleep window should be. The sleep window is going to be shorter than you want. That is where the real work begins, and the diary is why you can trust the starting point.

If your sleep is affecting daily life, talking to a GP or sleep specialist is a good first step. The American Academy of Sleep Medicine has a provider finder at sleepeducation.org. If insomnia is accompanied by depression, anxiety or other concerns, those are worth raising with your doctor as part of the same conversation.