Many adults on long-term sleep medication (zopiclone, zolpidem, lorazepam, temazepam, clonazepam, and others) reach a point where they would like to come off. The medication may have stopped working as well as it used to. Side effects may be accumulating. Guidance from a physician, a partner, or the person themselves may have made the case for a change. But the moment they try to stop, insomnia comes screaming back, often worse than before, and they conclude that they cannot come off.

The problem is usually not that the person cannot live without the medication. The problem is that abrupt discontinuation reliably produces rebound insomnia (and, for benzodiazepines, withdrawal), and this rebound is often mistaken for the underlying insomnia returning. The current evidence-based approach is different: a gradual, structured taper with CBT-I in parallel. Multiple randomized trials show this approach produces meaningfully better outcomes than either taper alone or CBT-I alone in medication users.

Here is what the current evidence says.

The evidence for CBT-I plus taper

The relevant clinical trials, going back to the seminal Morin et al. 2004 study, consistently show:

  • CBT-I plus supervised taper outperforms taper alone for successful discontinuation and for sleep quality maintenance after discontinuation.
  • Gains persist at follow-up. Morin's 3-month follow-up data showed sustained medication cessation and improved sleep efficiency.
  • Self-help CBT-I (Belleville et al., 2007) also outperformed taper alone, suggesting the effect is not just about therapist time.
  • CBT-I supports maintenance of adequate sleep during tapering, which is often the deciding factor in whether people can stick with the process.

The 2025 Alliance for Sleep clinical practice guideline and the 2025 ASAM/ACMT joint benzodiazepine tapering guideline both recommend CBT-I in parallel with a gradual taper as the current standard of care.

What a taper actually looks like

There is no one-size-fits-all schedule, but the principles are consistent across guidelines:

Gradual reductions. Most protocols reduce the dose by 10 to 25 percent every 1 to 4 weeks. Slower tapers are almost always better tolerated than faster ones, particularly for benzodiazepines and long-term Z-drug use. There is no medal for tapering fast.

Behavioural anchor. CBT-I is initiated before or in parallel with the taper, so that sleep restriction, stimulus control, and cognitive work are already building the non-medication sleep pattern as the medication is coming down.

Consistent wake time and time in bed. Non-negotiable, throughout the taper. This anchors the circadian system and builds sleep drive during a period when the medication is doing less to force sleep.

Managing rebound. Expect that some nights during the first two to four weeks of the taper will be worse than baseline. This is rebound, not a return of the "real" insomnia. Understanding this in advance is a big part of what makes people able to stick with the taper.

Cross-tapering, in some cases. For some benzodiazepines, particularly short-acting ones, switching to a longer-acting equivalent (diazepam) and tapering from there can reduce inter-dose withdrawal. This is a decision for the prescribing physician.

No abrupt stops. Especially not with benzodiazepines. Abrupt discontinuation of high-dose or long-term benzodiazepine use can produce seizures and is medically dangerous. Even Z-drugs (zopiclone, zolpidem, eszopiclone) should be tapered rather than stopped abruptly if use has been prolonged or supratherapeutic.

Which medications need what

Benzodiazepines (lorazepam, clonazepam, temazepam, diazepam, alprazolam). Slowest tapers, medical supervision essential, CBT-I in parallel. Withdrawal is real and can be significant.

Z-drugs (zopiclone, zolpidem, eszopiclone, zaleplon). Gradual taper recommended, especially for prolonged or supratherapeutic use. CBT-I in parallel is well-evidenced. Rebound insomnia is common.

Newer dual orexin receptor antagonists (daridorexant, lemborexant, suvorexant). Do not require the same slow taper protocol, but CBT-I still improves outcomes on discontinuation.

Antihistamines (diphenhydramine, doxylamine), over-the-counter sleep aids. No withdrawal syndrome, but tolerance and rebound insomnia are common. CBT-I is the appropriate replacement.

Trazodone, mirtazapine, low-dose doxepin (off-label for sleep). Should be reduced gradually with prescriber guidance. Discontinuation effects vary.

Cannabis for sleep. No formal taper protocol, but abrupt cessation after regular use commonly produces significant sleep disruption, vivid dreams, and rebound insomnia for one to three weeks. CBT-I in parallel is helpful.

What CBT-I contributes to the taper

CBT-I during the taper does three things simultaneously:

1. Builds the non-pharmacological sleep pattern. Sleep restriction and stimulus control raise sleep drive and consolidate sleep, which partially compensates for the medication effect being reduced.

2. Reframes rebound. Cognitive work on beliefs about the medication ("I can't sleep without it") and about the withdrawal ("this proves I still need it") is essential. Without this, the first few rough nights of the taper are often what causes people to abandon the attempt.

3. Builds durable change. By the time the medication is off, the sleep pattern has changed, so the underlying insomnia (the reason the medication was prescribed in the first place) is now being treated at its actual mechanism.

What to expect

The first two to four weeks of a taper are usually the hardest. Sleep is often worse than baseline. Rebound insomnia peaks in the first week or two after each dose reduction and then eases. By weeks 4 to 8, most people are noticeably below their starting dose and sleep is beginning to stabilize on the new lower level.

By the end of a properly-paced taper (often 8 to 20 weeks depending on the starting dose, medication class, and duration of use), most people are either off the medication or on a much lower dose, and sleep is often better than it was on the medication in the last year or two before starting the taper.

There are exceptions. Some people, particularly older adults on very long-term benzodiazepine use, may not be candidates for full discontinuation, and the appropriate goal may be dose reduction to the lowest effective level. These decisions are made in coordination with the prescribing physician.

Important safety notes

  • Do not taper without medical guidance. Especially not for benzodiazepines. The prescriber should be involved in setting the schedule and monitoring the process.
  • Do not taper during a major stressor (bereavement, job loss, other acute crisis) if it can be avoided.
  • Pause or slow the taper if symptoms of withdrawal become significant, or if depression, anxiety, or suicidality worsen.
  • Some medications need psychiatric input for taper (particularly benzodiazepines with mood disorder or PTSD indications, or where anxiety symptoms are prominent).

Where to start

If you are on chronic sleep medication and would like to come off, the appropriate first step is a conversation with your prescribing physician about the plan, coupled with a referral to a clinician trained in CBT-I who can support the process in parallel. The evidence strongly supports doing both together rather than either alone.

References

  • Alliance for Sleep Clinical Practice Guideline on switching or discontinuing insomnia medications, 2023.
  • American Society of Addiction Medicine and American College of Medical Toxicology. Joint clinical practice guideline on benzodiazepine tapering, 2025.
  • Morin CM et al. Randomized clinical trial of supervised tapering and cognitive behaviour therapy to facilitate benzodiazepine discontinuation in older adults with chronic insomnia. Am J Psychiatry, 2004.
  • Belleville G et al. Self-help CBT-I as an adjunct to medication tapering. Sleep, 2007.
  • Coteur K et al. CBT-I plus taper vs treatment as usual. 2022.
  • Lui YW et al. Long-term CBT-I plus taper outcomes. 2021.