GUIDE · SLEEP

Sleep restriction therapy: fewer hours in bed, and why a clinician usually guides it

You're in bed nine hours and asleep for six, and the nights still feel long. Sleep restriction therapy, one part of CBT-I, cuts your time in bed to about the hours you actually sleep, then lengthens it as your sleep firms up. It has trial support and real side effects. This page covers how the window is set, what the first weeks feel like, who shouldn't try it, and why it's usually run with a clinician.

What sleep restriction therapy is and where it comes from

Sleep restriction therapy (SRT) is a behavioral treatment for chronic insomnia. You're given a fixed sleep window, the hours you're allowed in bed, set close to the hours you actually sleep, and it lengthens as your sleep firms up. SRT usually sits inside a full program, which our CBT-I guide covers.

Arthur Spielman and colleagues Saskin and Thorpy described it in Sleep in 1987, starting from the idea that too much time in bed helps keep insomnia going. Thirty-five patients had their time for sleep sharply restricted, then extended as sleep efficiency improved, over eight weeks. They reported better sleep, and the gains held at a mean of 36 weeks in the 23 who came back for follow-up. With no comparison group, that study can't show the restriction caused the gains.

The idea rests on sleep pressure, the drive to sleep that builds while you're awake. Spielman and coauthors write that limiting the sleep you accumulate is one of the most reliable ways to strengthen it. That's the developers' explanation, and it doesn't predict how any one person will respond.

Who shouldn't try this, and why a clinician usually runs it

No source we read calls SRT safe for everyone. Don't start on your own if any of these applies, and see a doctor first if you can't tell.

• Dangerous drowsiness, or heavy sleepiness already: the AASM names heavy machinery operators, drivers, and people with excessive daytime sleepiness. Spielman's chapter adds air traffic controllers and long-haul truck and bus drivers, and Kyle's team suggests a more liberal window for people who are excessively sleepy at the outset.
• Bipolar disorder or a tendency toward mania: the AASM lists people predisposed to mania or hypomania. In a 2013 case series of 15 people with bipolar disorder, two of the five who had sleep restriction reported mild hypomania. The authors judged it safe in that small series and advised careful monitoring, which is a job for the clinician treating the bipolar disorder.
• Seizure disorders, sleep apnea, and parasomnias: the AASM cautions about poorly controlled seizure disorders, and Spielman's chapter says people with epilepsy, sleep-related breathing problems, or parasomnias shouldn't do SRT. The HABIT trial of nurse-delivered SRT also screened out restless legs syndrome and narcolepsy.
• Other groups HABIT excluded, among them: pregnancy or planning one within six months, night, evening, early-morning, or rotating shifts, schizophrenia, and current suicidal thoughts with intent or an attempt in the past two months. Its protocol calls these conditions that may make SRT inappropriate or ineffective.

The AASM's reasoning for recommending SRT assumes a clinician is monitoring you. A clinician can screen for these conditions, set the starting window with you, choose 15- or 30-minute steps, and catch sleepiness or mood problems early. HABIT used four nurse-led sessions.

How the sleep window is set and adjusted

The sources we read agree on the outline and differ on the numbers.

• Diary, then window: one to two weeks of diary with nothing changed gives your average sleep time and your sleep efficiency, which is time asleep divided by time in bed. The starting window is that average. Wake time stays the same every day, and bedtime is counted back from it.
• The floor: As Spielman's 2011 chapter describes it, the 1987 protocol gave no one less than 4 hours 30 minutes in bed. The chapter itself says 5 hours, as do the Kyle and Vallières studies. Stanford Health Care says 5.5 even if you sleep less, and so does our CBT-I guide. The higher floor is the cautious one.
• Weekly review: seven days of sleep efficiency decide the next change. Spielman's chapter and the Kyle study add 15 minutes at 90 percent or more, hold from 85 to under 90, and cut 15 below 85 (the chapter allows 15 or 30 either way). The Vallières study and Stanford add time at 85 or more, hold from 80 to 85, and cut below 80. Our guide uses the second set.

A worked example (invented numbers)

This example is invented: Dana is fictional, and her numbers aren't data.

Dana's diary averages 8 hours 15 minutes in bed and 6 hours 15 minutes asleep, about 76 percent efficiency. With a 6:30 a.m. wake time, a 6-hour-15-minute window puts her bedtime at 12:15 a.m.

In week one she sleeps 5 hours 30 minutes of it, or 88 percent. Under the second set of cutoffs she gains 15 minutes, and bedtime moves to midnight. In week two she sleeps 5 hours 15 minutes of 6 hours 30, about 81 percent, so nothing changes. The first set would have held her in week one and cut 15 minutes after week two. Ask which set your program uses.

What the first weeks feel like

Expect sleepiness, and possibly fog and irritability. The AASM names increased daytime sleepiness and trouble concentrating early in SRT, lists irritability among CBT-I's early effects, and says they typically fade. Spielman's chapter adds fatigue and memory problems, and notes that some people can't tolerate them long enough to benefit.

In a 2014 sleep-lab study of 16 adults, Kyle and colleagues measured sleep falling by an average of 91 minutes on night 1, versus baseline. Reaction-test lapses rose, and Epworth Sleepiness Scale scores went from 4.95 at baseline to 9.08 in week 2, returning to baseline by three months. Self-reported sleep and insomnia severity improved over the treatment.

A small 2018 study of 16 adults found no significant change in sleepiness or driving-simulator performance over two weeks, though its authors say that applies only when total sleep time stays about the same. You can't know which pattern you'll have, so plan for drowsiness.

How it differs from going to bed later or sleeping less

Going to bed later, or cutting sleep to toughen up, skips what makes SRT workable: a diary-based window, a fixed wake time, a floor, and weekly rule-based changes that widen the window as sleep firms up.

A 2021 meta-analysis of eight randomized trials found large effects on insomnia severity and sleep efficiency and none on total sleep time, so the aim is denser sleep.

How it pairs with stimulus control, and gentler variants

Clinicians usually teach SRT with stimulus control, the rules for leaving bed when you're awake and keeping it for sleep. Restriction sets how long you're in bed, and stimulus control covers what to do when sleep doesn't come.

Gentler starts exist. Spielman's chapter describes adding 30 minutes to the average sleep time or allowing a nap, and sleep compression shrinks time in bed in stages instead. A 2025 trial of 234 adults found it lowered insomnia severity less and more slowly than restriction, with better adherence and somewhat fewer side effects. The authors call it a valid alternative when restriction can't be used. We found nothing saying the cautions above don't apply, so treat them as applying.

What the research does and doesn't say

The AASM's 2021 guideline gives SRT a conditional, or weaker-tier, recommendation as a single-component therapy and strongly recommends only multicomponent CBT-I. It rates the evidence for SRT low, citing imprecision and risk of bias. In the 2021 meta-analysis, six of eight trials had a high risk of bias, and three or fewer reported follow-up outcomes.

One large trial, HABIT, enrolled 642 adults in English general practices. Four nurse-delivered sessions plus a sleep hygiene booklet beat the booklet alone on self-reported insomnia severity at six months (10.9 versus 13.9 on the Insomnia Severity Index). Eight people in each group had serious adverse events, none judged related to treatment. It excluded several of the groups above and wasn't blinded. Among the trials we read, none tested people starting alone from an article like this one.

How a program is run, in order

This is how clinicians run a sleep restriction program, so you can follow along with yours or know what CBT-I asks of you. Starting alone is an option only if none of the cautions above apply and you aren't already unusually sleepy. If you're unsure, see a doctor first.

1. Check the cautions before anything else. If anything in the list above applies or you can't tell, see your doctor first, and ask for a clinician who does CBT-I.
2. Keep a sleep diary for one to two weeks. Each morning, log bedtime, time to fall asleep, time awake at night, final wake-up, and time out of bed. Don't check the clock overnight, and change nothing yet.
3. Work out average sleep time and sleep efficiency. Average the hours you actually slept. Efficiency is time asleep divided by time in bed, times 100. Spielman's chapter gives under 85 percent (80 for older adults) as the usual reason to use SRT.
4. Set the window, and don't go below the floor. Your window is your average sleep time, with 5.5 hours as the minimum even if you sleep less. Pick a wake time you can keep every day and count back for bedtime. Stay out of bed outside the window.
5. Plan the first two weeks around drowsiness, and skip naps. Don't drive or do hazardous work when you feel drowsy. If sleepiness gets severe, lengthen the window and tell a doctor (Spielman's chapter describes adding 30 minutes to ease it). If you ever nod off at the wheel, stop the program and see a doctor. Skip naps unless a clinician prescribes one.
6. Review once a week and adjust by the rules. Average the past seven days. At 85 percent or more, add 15 minutes by moving bedtime earlier. From 80 to 85, hold. Below 80, cut 15 minutes, never below the floor. A clinician may use the other cutoffs above, so follow one set, not a mix.
7. Stop and get help if trouble shows up. Call a doctor if sleepiness becomes unmanageable, your mood swings sharply, you have a seizure, or you notice unusual behavior at night. Expect weeks, and treatment ends when the window supports good daytime functioning (Spielman's chapter).

If your nights are mostly racing thoughts, start with our guide to anxiety at night. The sleep test gives you a baseline score to compare against after a few weeks.

Check your diary arithmetic with Helpy

The chat below runs with Helpy, the site's AI assistant. Paste a week of diary times and it can work out your average sleep time and sleep efficiency, or go through the cautions list with you. Whether restriction is safe for you is a question for a doctor.

Evidence and sources

• Spielman, Saskin and Thorpy (1987), Sleep: the original report, 35 patients. Abstract only, uncontrolled.
• Spielman, Yang and Glovinsky (2011), Behavioral Treatments for Sleep Disorders: protocol numbers, contraindications, variants. A clinicians' chapter, not a trial.
• Stanford Health Care, CBTI: Sleep Restriction: the 5.5-hour floor and 85/80 rule. A clinic overview.
• Edinger and colleagues (2021), AASM guideline, J Clin Sleep Med: recommendation strengths, harms, cautions. Evidence rated low.
• Kyle and colleagues (2014), Sleep: sleepiness and vigilance in 16 adults, plus the protocol. Small, uncontrolled.
• Whittall and colleagues (2018), Sleep Medicine: 16 adults. Abstract only, small, uncontrolled.
• Vallières and colleagues (2013), Sleep Disorders: the 5-hour floor and 85/80 rule in use. A mechanism study.
• Kyle and colleagues (2023), HABIT trial, Lancet: 642 adults. Abstract only, open-label, self-reported outcome.
• Kyle and colleagues (2020), HABIT protocol, BMJ Open: the exclusion list. A protocol, not results.
• Kaplan and Harvey (2013), Am J Psychiatry: bipolar disorder. Abstract only, small case series.
• Maurer and colleagues (2021), Sleep Medicine Reviews: eight randomized trials. Abstract only, mostly high risk of bias.
• Jernelöv and colleagues (2025), Sleep: compression versus restriction, 234 adults. Abstract only.

Read next

How to beat insomnia without pills: CBT-I, step by stepThe full program this technique belongs toStimulus control for insomniaThe companion technique: give the bed its job backSleep tracker anxiety: when a score decides how tired you areWhen tracking the night makes it worseSleep hygiene: which habits help, and why they rarely fix insomnia on their ownWhich sleep habits have evidence behind them

← All CBT guides