Sleep

CBT for insomnia, self-directed

Cognitive behavioural therapy for insomnia (CBT-I) is the recommended first treatment for long-term insomnia. Online programmes get effects close to face-to-face treatment, though programmes with personal support, and four or more sessions with a therapist, do somewhat better. It is a method with specific rules, not a list of sleep tips, and it mostly fixes broken sleep and sleep-related worry rather than adding hours.

Evidence AStrong Last reviewed 7 min read

At a glance

  • The core: a fixed wake time every day, a sleep window set from your own sleep diary, and getting out of bed when you are lying awake.
  • What the studies show: across 87 trials, CBT-I had a large effect on insomnia severity compared with no treatment (g = 0.98) and improved sleep efficiency and time awake in the night; the effect on total sleep time was the smallest (g = 0.16).2
  • Expect: the first one to two weeks often feel worse before sleep consolidates. Plan for daytime sleepiness and do not drive drowsy.
  • Talk to a clinician first if you may have sleep apnoea, have bipolar disorder, take sleeping pills, or low mood or anxiety is behind the insomnia.

What the research shows

The American Academy of Sleep Medicine gives multicomponent CBT-I a strong recommendation as the first-line treatment for chronic insomnia in adults. The same guideline conditionally suggests stimulus control, sleep restriction or relaxation therapy as single treatments, and conditionally suggests against sleep hygiene advice on its own.1 Sleep hygiene, such as a dark room and no late coffee, is not what drives the benefit; the behavioural rules below are.

A meta-analysis of 87 trials (3,724 patients), compared with no treatment, found large effects on insomnia severity (g = 0.98), and medium effects on sleep efficiency (g = 0.71), time awake after falling asleep (g = 0.63) and time to fall asleep (g = 0.57). The smallest effect was on total sleep time (g = 0.16).2 CBT-I mainly makes sleep more continuous and less worrying, and does not mainly add hours. The same analysis found that face-to-face treatment of four or more sessions did better than self-help formats or very brief treatment.

Online programmes work too. In 11 trials (1,460 people), internet-delivered CBT-I had effects comparable to face-to-face treatment, generally kept up at 4 to 48 weeks, and the effect was larger when the programme included more personal support.3 If you can choose, pick a programme that gives you some contact with a person.

22:0000:0002:0004:0006:0008:00 Before: long time in bed, broken sleep First weeks: shorter window, same wake time Later: window widened by 15–20 min a week Fixed wake time Schematic of the method, not measured values
How the method works: a shorter window at first, with the same wake time every day, then a window that widens as sleep consolidates. Schematic: it shows the method, not measured values.

How to do it

This is for adults with chronic insomnia: trouble falling or staying asleep at least three nights a week for at least three months. The standard route is a structured 6–8 week programme (an app or website) that builds in the diary and the weekly adjustments below. If the full method feels like too much, the gentler version is a fixed wake time plus stimulus control (steps 2 and 4), without shortening the window (the cautions below still apply, because stimulus control also costs some sleep at first).

  1. Keep a sleep diary for 1–2 weeks. Each morning write down when you went to bed, about how long it took to fall asleep, how many times you woke, when you woke for good and when you got up. Work out your average hours of actual sleep.
  2. Fix your wake time. The same time every day, weekends included.
  3. Set your sleep window from the diary. Make it close to your average sleep time, never less than about 5–6 hours, and count back from your wake time to find the bedtime. Do not go to bed before it.
  4. Use the bed only for sleep (and sex). Go to bed when sleepy, not just tired. If you do not fall asleep within about 20 minutes (estimate it, do not watch the clock), get up, do something quiet in dim light, and go back when sleepy. Repeat as often as needed.
  5. Adjust once a week. When sleep is consistently efficient (you fall asleep fairly quickly and are asleep for nearly all the time you spend in bed; programmes often use about 85–90% as the line), widen the window by 15–20 minutes. If efficiency drops, hold the window or tighten it.
  6. Work on the worry too. Notice thoughts such as "I will be wrecked tomorrow" and practise answering them calmly. CBT-I includes this because worry about sleep keeps insomnia going.
  7. Stay with it. The first one to two weeks of a tighter window often feel worse before sleep consolidates, and daytime sleepiness usually starts in the first week.

An example start

Your diary shows about 5.5 hours of sleep on average, in bed from 23:00 to 07:00. That is 5.5 hours asleep out of 8 in bed, an efficiency of about 69%. You choose a fixed wake time of 06:30 and a window of 6 hours (not less than 5–6 hours), so bedtime is 00:30.

WeekWhat you do
Before startingKeep the diary for 1–2 weeks. No changes yet.
Week 1Up at 06:30 every day. In bed from 00:30. Out of bed if awake for about 20 minutes. Keep the diary. Expect to feel sleepy in the day.
Week 2Same window. No driving or risky tasks if drowsy.
Week 3 onwardIf sleep is consistently efficient, move bedtime 15–20 minutes earlier (00:10–00:15). Repeat weekly. If efficiency drops, hold or go back.

Who should be careful

Common mistakes

  • Spending longer in bed to "catch up". This is the opposite of what the method does and can deepen insomnia.
  • Giving up after a rough first week.
  • Skipping the diary. It is what lets you set the window and adjust it each week.
  • Treating CBT-I as sleep hygiene tips only. Hygiene is the weakest part; the restriction and stimulus-control rules carry most of the effect.
  • Watching the clock. Estimate the 20 minutes instead.
  • Long or late naps, which use up the sleepiness the method relies on. See the sleep regularity guide for a steady wake time and naps.

How to tell it is working

Fill in the Insomnia Severity Index (a short questionnaire) at the start and every 2–4 weeks; a falling score is the main sign. The diary shows the rest: falling asleep faster, fewer or shorter wake-ups and more of your time in bed spent asleep. Total sleep time may change little, and that is expected.

How strong is the evidence

Grade A. A major guideline gives CBT-I a strong recommendation, and large meta-analyses show big effects on insomnia severity compared with no treatment, including for internet-delivered programmes. The honest limits: the trials mostly enrolled adults without an untreated sleep disorder such as apnoea, self-help formats do a little worse than four or more sessions with a therapist, and the effect on total sleep time is small.

References

Each reference links to PubMed or the publisher. The tag shows what kind of study it is; the grade above reflects all of them together, not any single one.

  1. Guideline Edinger JD, et al. Behavioral and psychological treatments for chronic insomnia disorder in adults: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med 2021. Strong recommendation for multicomponent CBT-I as the first-line treatment for chronic insomnia; it conditionally suggests against using sleep hygiene advice on its own. PubMed 33164742 · DOI
  2. Meta-analysis van Straten A, et al. Cognitive and behavioral therapies in the treatment of insomnia: a meta-analysis. Sleep Med Rev 2018. 87 trials (3,724 patients) compared with no treatment: a large effect on insomnia severity but the smallest effect on total sleep time; face-to-face treatment of four or more sessions beat self-help. PubMed 28392168 · DOI
  3. Meta-analysis Zachariae R, et al. Efficacy of internet-delivered cognitive-behavioral therapy for insomnia: a systematic review and meta-analysis of randomized controlled trials. Sleep Med Rev 2016. 11 trials (1,460 people): online CBT-I worked about as well as face-to-face treatment, and more personal support went with bigger effects. PubMed 26615572 · DOI
How I grade the evidence

Every guide carries a grade. It describes how well the research supports the advice, not how important the advice is. Grades come from a written rubric and are re-checked as new studies appear.

  • AStrongConsistent results from high-quality meta-analyses or several trials, a clear size of benefit, well-understood risks.
  • BModerateAt least one good systematic review or several solid trials. Promising, with real uncertainty about dose, groups or the long term.
  • CLimited but plausibleSome human evidence, but small, mixed or specific to one condition. Reasonable and low-risk, not proven.

Grade D (animal, mechanistic or influencer-driven claims) and E (disproven or unsafe) are not published here.

Want this fitted to you?

A consultation turns general evidence into a short plan for your history, results and life.

Request a consultation