🌙 Quitude

Learn › CBT-I

CBT-I vs sleeping pills: what the evidence actually says

Sleeping pills and CBT-I (cognitive behavioural therapy for insomnia) can both help you sleep, but they behave differently once treatment stops. Medication works while you take it. CBT-I teaches skills that, for many people, keep working after the programme ends. That difference is a large part of why sleep medicine guidelines put CBT-I first for chronic insomnia and treat medication as a shorter-term or fallback option. Neither is a magic switch and neither is a cure, and there are people for whom medication is genuinely the right call. This page lays out both sides plainly, including the parts of the comparison that are less flattering to CBT-I than the usual telling.

The short version: speed versus durability

Sleeping pills act on the brain's arousal and sleep-regulating systems for the hours the drug is active. That is their strength: they can work the first night, which matters enormously if you are three weeks into a bad patch and running on empty. It is also their limit. When the drug leaves your system, the underlying insomnia is usually still there.

CBT-I works on the things that keep insomnia going after whatever originally started it has passed. Chronic insomnia is typically maintained by two habits that feel entirely sensible from the inside: spending more time in bed to 'catch up', which dilutes your sleep across too many hours, and the bed slowly becoming a cue for lying awake, checking the clock and bracing for another bad night. CBT-I addresses both directly. Sleep restriction matches your time in bed to how much you are actually sleeping, so sleep consolidates, and stimulus control re-teaches your body that bed means sleep. Sleep restriction is not suitable for everyone without medical advice, so before you try it, read the safety section below on who should speak to a clinician first. Cognitive work on catastrophic sleep thoughts and counter-arousal techniques sit alongside those two.

This is worth being blunt about: CBT-I is not a relaxation programme, and it is not sleep hygiene. The American Academy of Sleep Medicine's 2021 guideline made a conditional recommendation against using sleep hygiene on its own as a treatment for chronic insomnia. Not because good habits are bad, but because on their own they are not enough.

What the guidelines actually say

Every major body that has looked at this puts behavioural treatment first, and it is worth seeing the actual wording rather than the summary.

The AASM's 2021 guideline on behavioural and psychological treatments for chronic insomnia in adults made six recommendations. Exactly one is graded STRONG: that clinicians use multicomponent CBT-I. The other five, covering brief behavioural therapies, stimulus control alone, sleep restriction alone, relaxation alone and sleep hygiene as a single component, are conditional.

The AASM's 2017 guideline on medications made fourteen recommendations, all of them graded WEAK. Weak in favour, versus no treatment, for suvorexant, eszopiclone, zaleplon, zolpidem, triazolam, temazepam, ramelteon and doxepin; weak against trazodone, tiagabine, diphenhydramine, melatonin, L-tryptophan and valerian for chronic insomnia in adults. On long-term use, that guideline says ongoing hypnotic use should be reserved for people for whom CBT is inaccessible or ineffective, who maintain long-term gains on medication, and who are followed up regularly.

The American College of Physicians reached the same conclusion from the primary-care side. Its 2016 guideline recommended CBT-I as the initial treatment for chronic insomnia in all adults, a strong recommendation on moderate-quality evidence, and only weakly recommended shared decision-making about short-term medication when CBT-I alone had not worked. It noted that these medications are FDA-approved only for short-term use, around four to five weeks, and that patients should not continue them for extended periods. That guideline is now past ACP's five-year review window, so treat it as a historical position rather than a live one.

In the UK, NICE's 2022 appraisal covered Sleepio, one specific digital CBT-I product rather than the category. It found reduced insomnia symptoms compared with sleep hygiene and with sleeping pills, and recommended Sleepio in primary care as a cost-saving option for people who would otherwise be offered those. NICE also noted there was no direct evidence comparing it with face-to-face CBT-I. That appraisal applies to Sleepio only. It does not extend to digital CBT-I apps generally, and it does not extend to Quitude.

The most recent piece is the AASM's 2026 clinical practice guideline on combination treatment, published in the Journal of Clinical Sleep Medicine in April. It makes a conditional recommendation for combining CBT-I with medication over medication alone, and a conditional recommendation against combination over CBT-I alone. Both are conditional and rest on low-certainty evidence. The first half matters if you are already taking a sleeping pill: current guidance conditionally favours adding CBT-I to what you are doing rather than treating it as a choice between the two.

How to read those grades (and what they do not mean)

It is tempting to line up one STRONG recommendation against fourteen WEAK ones and call that the case closed. The guideline's own authors tell you not to. AASM 2017 states plainly that a weak recommendation should not be construed as an indication of ineffectiveness, that GRADE strengths refer to the strength of the published evidence rather than the size of the effect in any particular patient, and that downgrading was predictable here given the funding source for most drug trials and the attendant risk of publication bias.

Three more things cut against the simple reading. That same 2017 guideline says hypnotic medications are comparably efficacious to CBT-I during acute treatment. CBT-I's strong 2021 recommendation rests on moderate-quality evidence, while triazolam and L-tryptophan were both rated high-quality evidence and still received weak recommendations, so the grades do not stack neatly. And the two guidelines came from different task forces answering different questions, which makes a direct grade-for-grade comparison shaky.

So the guideline hierarchy is a real signal but a secondary one. The stronger signal is what happens to people after treatment ends, which is the next section.

What happens after treatment stops

The more useful comparison is not 'which works tonight' but 'where are you six months from now'. A systematic review and network meta-analysis by Furukawa and colleagues looked at exactly that, with a median follow-up of about 24 weeks.

Across 9 trials and 627 people with long-term follow-up data, the model estimated long-term remission at about 41% for CBT-I (95% confidence interval 31 to 53%), compared with about 28% for medication: an odds ratio of 1.82, rated high certainty. Combination treatment came in around 40%, showing no clear advantage over CBT-I alone. These are model-derived estimates from a modest evidence base rather than plain observed rates in one large trial, and the confidence interval is worth keeping in view.

The dropout numbers point the same way and are often overlooked. Around 21% of people dropped out of CBT-I versus around 39% of those on medication. That is the opposite of the usual assumption that pills are the easy option and therapy is the one people quit.

None of this means everyone who completes CBT-I sleeps well afterwards. Remission at 41% also means most people in that arm were not in remission at follow-up. Remission is a strict all-or-nothing threshold, though, and plenty of people who do not quite reach it still sleep meaningfully better than when they started. Insomnia is stubborn, and some people need a second round, a different emphasis, or a clinician's help.

Stopping a sleeping pill is its own medical event

The guidelines' caution about long-term hypnotic use reflects well-known issues with sustained use: next-day grogginess, the drug becoming less helpful over time for some people, and the difficulty of stopping once sleep and the pill have become psychologically linked. Many people describe not being physically dependent so much as being unable to face a night without the safety net.

Here is the part that gets softened too often. If you are already on a prescribed sleeping pill, do not stop or cut it down on your own on the strength of an article like this one. The FDA requires a boxed warning that abruptly stopping a benzodiazepine after weeks or months of use can cause serious, life-threatening withdrawal reactions, including seizures. Withdrawal seizures are also documented with zolpidem. Temazepam and triazolam, both named in the guideline list above, are benzodiazepines. Stopping suddenly is not just a rough week of sleep: it can mean rebound insomnia, anxiety and tremor, and with benzodiazepines and some other hypnotics, seizures. Any change to a prescribed sleeping pill needs to be planned and supervised by your prescriber.

It is also worth knowing that several things people reach for first, precisely because they seem gentler than a prescription, did not fare well in the 2017 AASM review. Diphenhydramine, the sedating antihistamine in most over-the-counter sleep aids, along with melatonin, valerian, L-tryptophan and trazodone, all received weak recommendations against for treating chronic insomnia in adults. Melatonin has other legitimate uses, such as jet lag and shift work, but the evidence did not support it as a treatment for chronic insomnia. Being available without a prescription is not the same as being effective for it.

The honest cost of CBT-I

CBT-I asks something of you that a pill does not. You keep a sleep diary for a week or two before anything is prescribed, because the programme sets your time in bed from your own data rather than a generic target. Then, for the first couple of weeks, you will likely spend less time in bed than you do now, and many people feel sleepier during the day before they feel better at night.

Some extra daytime sleepiness in the first weeks is expected, and it is not a sign of failure. It is a real safety issue while it lasts, though. Do not drive or operate machinery when you feel too sleepy to do it safely, and if the sleepiness becomes severe, or you find yourself struggling to stay awake during the day, add time back to your sleep window and speak to a clinician rather than pushing on. The section below sets out who should not start sleep restriction without medical advice at all.

Good programmes also put a floor under how far time in bed can be cut, and widen the window again as sleep efficiency improves. Quitude never prescribes a window shorter than 5 hours 30 minutes. That floor was raised from 5 hours precisely because Quitude is unsupervised self-help, with no clinician checking on you week to week. It reviews your window weekly using 7 days of diary data and Spielman's convention: sleep efficiency of 90% or above widens the window by 15 minutes, 85 to 90% holds it steady, and below 85% trims it by 15 minutes, never below the floor. It applies the same thresholds to everyone rather than age-stratified variants.

Quitude's seven-week programme sits inside the usual four-to-eight-session range and is free. It delivers the full protocol, meaning the diary, a personalised sleep window, weekly titration, stimulus control and the cognitive tools, rather than a layer of sleep-hygiene tips. It is self-help software following the published clinical protocol, and it has not itself been tested in a clinical trial. The trial evidence discussed on this page comes from clinician-delivered CBT-I and from separately tested digital products such as Sleepio and Somryst. The method is the thing with the evidence behind it.

Before you narrow your sleep window: who should check first

Sleep restriction deliberately reduces your sleep opportunity for a while, and that has real consequences. Talk to a clinician before you start, not after, if any of the following apply to you.

This next part applies to everyone on the programme, not only people who drive for a living: do not drive or operate machinery when you feel too sleepy to do it safely. Sleepiness affects reaction time before it feels dramatic, and the early weeks are exactly when it peaks. Plan the first fortnight around that, including lifts, public transport or a later start if you can arrange one.

Safety outranks the protocol. The programme asks you to skip naps so sleep pressure builds, but if you are too sleepy to be safe, especially before driving, a short nap early in the day is the right call. Treat it as information rather than a failure: it usually means your window needs widening or your body needs a clinician's eyes on it.

Stop and get advice, rather than pushing through, if you are falling asleep unintentionally during the day, nodding off while driving or at work, or if your mood becomes elevated, racy or agitated, or drops significantly. Add time back to your sleep window and speak to a clinician. None of those are the programme working as intended. If your mood drops badly, or you feel hopeless or unsafe, get help the same day: in the US you can call or text 988, and elsewhere contact your local emergency services or crisis line.

This page is general information about treatment options, not medical advice, and it cannot tell you what is right for your situation.

  • A seizure disorder, particularly one that is poorly controlled: sleep loss can lower the seizure threshold
  • Bipolar disorder, or a history of mania or hypomania, where sleep loss can trigger an episode
  • Untreated or suspected obstructive sleep apnoea: loud snoring, witnessed breathing pauses, or waking unrefreshed however long you sleep
  • A parasomnia such as sleepwalking, night terrors or acting out dreams, which sleep loss is known to trigger
  • Narcolepsy, or any condition already causing excessive daytime sleepiness
  • Safety-critical work: driving, machinery, healthcare or shift roles where a lapse in attention is dangerous
  • Pregnancy, or another significant medical or psychiatric condition
  • A raised risk of falls, which matters most for older or frail readers moving around a dark house while sleepy

Who each option actually suits

Framing this as a moral contest, pills bad and therapy good, is unhelpful and does not match what clinicians actually do. The honest split looks more like this.

Where to start tonight

If you are reading this at 2am, the most useful thing is not a decision about medication. It is a week of honest sleep diary data, because everything in CBT-I is built from it. You do not have to change anything yet.

Fill the diary in the next morning, not during the night. Rough estimates are exactly what the programme needs, and checking the clock at 3am to make the numbers accurate works against you, since clock-watching is one of the habits CBT-I later works to undo. So if you are awake right now, the diary can wait until morning. Then record roughly when you got into bed, roughly when you fell asleep, roughly how long you were awake in the night, and when you got up.

Before you change your schedule at all, check the safety list above. If you have a seizure disorder, bipolar disorder, a parasomnia, possible untreated sleep apnoea, narcolepsy or existing daytime sleepiness, a safety-critical job, or a raised risk of falls, speak to a clinician before restricting your time in bed.

If your insomnia has lasted more than three months, is affecting your days, or comes with symptoms that suggest another sleep disorder or a low mood you cannot shift, see a clinician. If you feel hopeless or unsafe, do not wait for an appointment: in the US, call or text 988, and elsewhere contact your local emergency services or crisis line.

Frequently asked questions

Can I start CBT-I while I'm still taking sleeping pills?

Yes. Many people do, and CBT-I does not require you to be medication-free to begin. The AASM's 2026 combination guideline conditionally favours CBT-I plus medication over medication alone, so adding the programme is a reasonable move. What matters is that you do not stop or reduce your medication on your own. Abruptly stopping a benzodiazepine after weeks or months can cause serious withdrawal reactions including seizures, and withdrawal seizures are documented with zolpidem too, so any change needs to be planned and supervised by your prescriber. For many people the medication becomes something to step down from over time, but that is a decision to make with the prescriber based on how you respond. Guidelines do support staying on medication for some people, particularly where CBT-I is not available or has not worked.

How fast does CBT-I work compared with a sleeping pill?

A sleeping pill can work the first night. CBT-I is typically four to eight sessions spread over four to eight weeks, and the first week or two often feel harder, not easier, because the sleep window is deliberately narrowed. Sleep restriction is not suitable for everyone without medical advice, so check the safety section above before you narrow anything. When it works, sleep tends to consolidate first, with fewer and shorter awakenings, and total sleep time builds back up as the window is widened week by week. The trade is speed for durability: at around six months of follow-up, remission rates favour CBT-I over medication.

Is melatonin a safer alternative to prescription sleeping pills?

For chronic insomnia in adults, the AASM's 2017 medication guideline made a weak recommendation against melatonin, because the evidence did not support it as a treatment for this condition. That is separate from its use for circadian problems such as jet lag or shift-work sleep disruption, where the reasoning is different. Several other over-the-counter options fared the same way in that review, including diphenhydramine, the sedating antihistamine in most drugstore sleep aids, and valerian. Being available without a prescription is not the same as being effective for chronic insomnia, and it is still worth telling your doctor what you are taking.

Does taking medication and doing CBT-I together work better than either alone?

It depends what you are comparing it with. Against medication alone, the AASM's 2026 guideline conditionally recommends combination treatment, on low-certainty evidence. Against CBT-I alone, the same guideline conditionally recommends against it, and the longer-term data agrees: in a network meta-analysis with a median follow-up of about 24 weeks, combination showed roughly 40% long-term remission versus roughly 41% for CBT-I alone, so no clear advantage. In practice, clinicians often start people on both when someone is exhausted and needs short-term relief while learning the skills, with a plan to taper the medication later rather than continue it indefinitely.

Are sleeping pills ever the right choice?

Yes. Guidelines do not say never; they say not first, and not indefinitely by default. Short-term, time-limited use through shared decision-making with a clinician is a reasonable option during an acute crisis, while waiting for access to a CBT-I programme, or when CBT-I has been genuinely tried and has not worked. The AASM's position is that ongoing hypnotic use should be reserved for people for whom CBT is inaccessible or ineffective, who maintain their gains on the medication, and who are screened and followed up regularly. For those people, staying on treatment is the guideline-concordant outcome, not a failure.

Try the program free

Quitude runs the CBT-I protocol from your own sleep diary — the program is free, there is no account, and your data stays on your device.

Start the CBT-I program

Keep reading