INSOMNIA — CBT-I PROTOCOL
Sleep restriction therapy schedule — seven worked protocols
The Spielman 1987 protocol is one rule. The seven examples below show what that rule looks like across the cases that actually walk into clinics.
The hard part of sleep restriction therapy isn't the rule — it's the first 10 days, the edge cases the rule book doesn't cover, and the moment you decide whether your week of bad sleep means the protocol is failing or working as designed.
These seven worked examples cover the cases that produce the most questions: severe insomnia, sleep-onset versus sleep-maintenance subtypes, shift work, anxiety pairing, and older adults. Each one ends with the rule you'd apply to your own log on day 7.
How to calculate YOUR starting window
Log seven consecutive nights. For each night record: time got into bed, estimated sleep onset latency (in minutes), number of awakenings + estimated total wake time, final wake time, time got out of bed. From that, derive total sleep time per night = TIB minus onset latency minus wake-after-sleep-onset.
Average the seven nights. That average TST becomes your prescribed TIB, with a floor of 5h30m for adults under 65 and 5h45m for adults 65+. Pick the wake-time first (anchored to your real obligations), then back-calculate bedtime. Hold the window every night for at least 7 nights before evaluating.
The walk-through with example numbers lives in the calculator itself. Open it now and run your own log alongside the examples below — the examples will make more sense in your own context.
Example 1: severe insomnia, TST 5h
Baseline: 8h in bed, ~5h of actual sleep, sleep efficiency ~63%.
Mary, 41. Wakes at 6:30am for work. Has been in bed 10:30pm to 6:30am (8h TIB) but logs 5h average TST across the past 7 nights — 60–90 min sleep onset, 1–2 awakenings of 20–30 min each.
Prescribed window: 5h30m (the adult floor — her measured TST of 5h is below floor, so the floor applies). Anchor wake at 6:30am. Therefore bedtime is 1:00am. She does not get in bed before 1:00am. She gets out of bed at 6:30am regardless of how the night went.
Week 1 will be hard. Daytime sleepiness around 2–4pm is expected and is the therapeutic signal — it means homeostatic pressure is accumulating. By night 4–5, sleep efficiency climbs into the 80s. By day 7 it usually exceeds 85% for five consecutive nights, and the window expands to 5h45m (1:00am → 6:45am, or 12:45am → 6:30am — preserve the anchored wake).
Example 2: moderate insomnia, TST 6h
Baseline: 8h in bed, ~6h of actual sleep, sleep efficiency ~75%.
James, 36. Wakes at 7:00am. In bed 11:00pm to 7:00am (8h TIB), logs 6h average TST — 30–45 min onset, occasional brief awakenings.
Prescribed window: 6h. Anchor wake at 7:00am. Bedtime moves to 1:00am.
Easier first week than Example 1 because the gap between TIB and TST was smaller to begin with. Onset latency typically drops within 4–6 nights. Once efficiency exceeds 85% for five consecutive nights, expand to 6h15m the following week, then 6h30m the next.
Example 3: sleep-onset insomnia, late bedtime drift
Baseline: in bed 11pm, asleep ~1am, up at 7am for work. TST ~6h.
Priya, 28. Onset latency of 60–120 minutes most nights; once she's asleep, sleep is continuous. Wakes fine on weekends if she sleeps later.
Anchor-wake approach: pick a real wake time first (7am, weekday-stable). Prescribed TIB equals measured TST (6h). Bedtime becomes 1am. She does not get into bed before 1am.
For sleep-onset insomnia specifically, the bedtime delay does most of the work. The arousal at bedtime is partly conditioned ("I lie in bed and can't sleep, so I associate bed with not sleeping"). Delaying entry into bed until she's actually sleepy retrains the association. Expansion proceeds 15 minutes earlier per week as efficiency stays above 85%.
Example 4: sleep-maintenance insomnia, fragmented nights
Baseline: in bed 10pm, asleep within 15 min, but 2–3 awakenings of 30–60 min each. TST ~5h30m.
David, 52. Falls asleep easily but wakes at 2am, 4am, and sometimes 5:30am. Cumulative wake-after-sleep-onset of 90–120 minutes. TIB 9h, TST 5h30m, efficiency ~61%.
Prescribed window: 5h30m (the adult floor). Anchor wake at 6:30am. Bedtime: 1:00am.
Sleep-maintenance pattern responds slightly slower than sleep-onset to SRT — the pressure has to overcome whatever's causing the mid-night arousals, not just initial onset difficulty. Pair with stimulus control (if awake more than 15–20 min, get out of bed) which is the strongest single CBT-I component for this subtype. The combined protocol usually consolidates within 2 weeks.
Example 5: shift worker, rotating schedule
Baseline: 3 nights, 3 days, 2 off — repeating. Sleep windows shift weekly.
Aisha, 34. Nursing schedule with rotating shifts. Modified SRT applies, because the standard rule assumes a stable schedule. Use the alertness-curve tool to map sleep opportunities to her actual rotation.
For her, the SRT window is calculated per shift block. Night-shift block (working 7pm-7am): target sleep window 8:30am-2:30pm (6h TIB after a one-hour wind-down). Day-shift block (7am-7pm): target window 9pm-3:30am, with a strategic 90-min nap before the shift if needed.
On days off, the rule is: don't expand the window to catch up. Maintain the prescribed TIB. Catch-up sleep on off-days teaches the system that off-day timing is sleep timing, which fragments the next shift's adaptation. The literature on shift work disorder is consistent on this — the calmest schedule wins, not the most-sleep schedule.
Example 6: insomnia + anxiety, cognitive component
Baseline: TST 5h, onset latency 60+ min, rumination dominant.
Carlos, 44. Diagnosed generalized anxiety. Sleep onset characterized by racing thoughts; even when physically tired, cognitive activation prevents the transition. TIB 8h, TST 5h, efficiency 63%.
Prescribed window: 5h30m (floor). Anchor wake at 6:30am, bedtime 1:00am. So far same as Example 1.
The pairing component: at bedtime, run cognitive shuffle (Beaudoin protocol) or paradoxical intention rather than waiting for sleep passively. The shuffle interrupts the verbal-rumination network that anxiety-driven insomnia depends on. Without the cognitive pairing, SRT alone produces about half the response in this subtype. With pairing, response rates approach the non-anxious case. The mind-anxiety hub covers the specific shuffle protocol.
Example 7: older adult, 60+
Baseline: TIB 8.5h, TST 5h45m, multiple brief awakenings, daytime drowsiness.
Margaret, 68. Lifelong good sleeper until two years ago. Now wakes 4–5 times per night; cumulative wake 60–90 min. Naps 30–45 min in the afternoon.
Prescribed window: 5h45m (the 65+ floor, slightly higher than the under-65 floor of 5h30m). Anchor wake at 6:30am, bedtime 12:45am. Naps allowed: brief (max 20 min, before 3pm) — the older-adult literature (Edinger 2005) supports this exception because total sleep needs shift with age but daytime function still matters.
Slower expansion: 15 min per week rather than 30, because older adults respond more gradually and the lower TIB floor leaves less room to overshoot. Watch for fall risk during week 1 daytime sleepiness — recommend not driving or doing hazardous tasks until the adaptation period is past.
When to expand the window
The rule: five consecutive nights with sleep efficiency ≥85%. On the morning of night 6, expand by 15–30 minutes. Younger adults can expand 30 min; older adults expand 15 min.
Add the time to bedtime, not wake time. Wake time stays anchored to your real schedule. Bedtime moves earlier across weeks as the window grows.
Plateau handling: if efficiency stays in the high 80s but never crosses 85% for 5 consecutive nights, you're at the floor of your individual sleep need. That's the target — hold the window. The goal is consolidated sleep, not maximum hours.
When SRT isn't working
After 4–6 weeks of strict protocol compliance with no measurable improvement, the assumption shifts. Either (a) the diagnosis is wrong (consider OSA, depression-related sleep disturbance, chronic short sleep rather than insomnia), or (b) something else is interfering (alcohol, late stimulants, undiagnosed medical condition).
The differential at this point: STOP-BANG screen for sleep apnea, depression screening (PHQ-9 ≥10), medication review (antidepressants that activate, decongestants, late steroids). If all clear, the protocol may need clinician-supervised intensification — moving the floor lower temporarily — which is outside the DIY scope.
And: not everyone needs SRT. If your TST baseline is already 7+ hours and you just dislike the quality of your sleep, SRT is the wrong tool. The wrong tool can make a marginal case worse. Re-read the criteria for SRT candidacy in the parent article before starting.
App-based protocols vs. DIY
DIY SRT, run from a paper log and the calculator on this site, has the same active ingredients as any branded digital CBT-I program: stimulus control, sleep restriction, cognitive component, sleep hygiene baseline. The difference is structure, accountability, and the cost of failure.
App-based programs (Sleepio is the most evidence-rich, Espie 2012 / 2019 RCTs) add: weekly progress check-ins, automated adjustment recommendations, sleep diary tracking that doesn't depend on you remembering to fill it in, and a clinician-developed cognitive component you can't easily reproduce on your own.
Worth paying for if: you've tried DIY for 2–3 weeks and stopped because the protocol was hard to track, or your case includes prominent cognitive symptoms where the structured cognitive component matters. Not worth it if: you're disciplined and your case is straightforward sleep-onset or sleep-maintenance without a strong cognitive overlay.
Sleepio is the program with the strongest RCT evidence base — Espie 2012, Espie 2019, and several follow-ups have shown response rates comparable to face-to-face CBT-I in the populations studied. UK NHS-approved. Worth considering if you've stalled on DIY for the structural reasons above.
The parent article on sleep restriction therapy covers the theory, candidacy criteria, and the cases where SRT is the wrong tool.
SRT pairs most strongly with stimulus control therapy — the second-strongest single CBT-I component, and the rule that handles the mid-night awakening edge case.
If your case carries a strong cognitive component, the protocol pairing referenced in Example 6 is detailed at anxiety insomnia.
Example 6's pairing technique — cognitive shuffle — is fully covered in mind racing at bedtime, the technique reference for the cognitive component.
If your SRT case overlaps with a same-hour 3am wake, the cortisol-loop piece pairs with this: 3am cortisol awakening.
Example 7's older-adult version connects to the broader insomnia in older adults piece — the disentanglement of normal aging from treatable insomnia disorder.
Example 5's rotating-shift case is covered in more depth in shift work disorder, including the harm-reduction protocol that fits around SRT.
If DIY SRT stalls, the structured app-based programs (Sleepio, Somryst, Sleep Reset) are compared in best CBT-I online programs.
On the choice between SRT and medication for chronic insomnia, CBT-I vs sleeping pills walks through the comparison.
The full set of CBT-I component articles lives at the insomnia hub.
FAQ
Are naps allowed during SRT?
For under-65 adults, no naps during the first 2–3 weeks. The whole point is to build homeostatic pressure; a 30-minute afternoon nap discharges roughly two hours of that pressure. For 65+, brief naps (under 20 minutes, before 3pm) are tolerable — the older-adult literature supports this exception.
What about weekends?
Same window, every day, including weekends. Especially weekends. Catch-up sleep on Saturday morning teaches the circadian system that 8am is now wake time, fragments the next week's adaptation, and is the single most common reason DIY protocols stall. The two-hour weekend drift is the enemy.
What if I get sick during week 1?
Real illness (fever, flu) is an exception — your sleep architecture changes during acute illness and forcing the window will not help recovery. Pause the protocol, sleep as needed, restart at the prescribed window when symptoms clear. Mild colds without fever are not an exception; keep going.
I missed two nights — do I restart from week 1?
No, restart from the day after the miss. Two missed nights costs you maybe 3–4 days of progress, not the whole 7-night counter. The cleanest path is acknowledge the slip and resume — perfectionism about the protocol is itself a form of sleep-related anxiety, and that anxiety is part of what SRT treats.
Sources
- Spielman AJ, Saskin P, Thorpy MJ. Treatment of chronic insomnia by restriction of time in bed. Sleep 1987.
- Edinger JD, Wohlgemuth WK, Radtke RA, Marsh GR, Quillian RE. Cognitive behavioral therapy for treatment of chronic primary insomnia: a randomized controlled trial. JAMA 2001.
- Edinger JD, Sampson WS. A primary care 'friendly' cognitive behavioral insomnia therapy. Sleep 2003 — practical clinical implementation guidance for SRT.
- Morin CM, Bootzin RR, Buysse DJ, Edinger JD, Espie CA, Lichstein KL. Psychological and behavioral treatment of insomnia: update of the recent evidence (1998-2004). Sleep 2006.
- Espie CA, Kyle SD, Williams C, Ong JC, Douglas NJ, Hames P, Brown JS. A randomized, placebo-controlled trial of online cognitive behavioral therapy for chronic insomnia disorder delivered via an automated media-rich web application. Sleep 2012 — Sleepio first major RCT.
- Sateia MJ, Buysse DJ, Krystal AD, Neubauer DN, Heald JL. Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults. American Academy of Sleep Medicine 2017 — context for CBT-I as first-line.