Coming off sleeping pills, with your doctor: what the research says

Studies of people who came off long-term sleeping pills found that more of them succeeded with a clinician-supervised, gradual reduction than with usual care. Adding CBT-I helped some outcomes, although not every study found a lasting extra benefit. Please do not stop or change a sleeping medicine on your own. Talk with your doctor or pharmacist first. ^[1][2]

This article describes research only. It does not tell you to stop, and it gives no doses, schedules or rates of reduction, because those must be set by your own prescriber for your own situation. For how the medicines compare with CBT-I in the first place, see CBT-I vs sleeping pills.

Please read this first

• Do not stop or change a sleeping medicine without your doctor or pharmacist.
• Stopping some of these medicines suddenly can be dangerous. A 2025 review describes withdrawal syndromes after sudden stopping of drugs such as benzodiazepines, and says they can include seizures and delirium. ^[8]
• Your prescriber decides whether, when and how to reduce a medicine. The Canadian guideline describes its own recommendations as meant to assist, not dictate, decision making between patient and clinician. ^[1]

Which medicines does this cover?

The Canadian deprescribing guideline covers benzodiazepines and the related "Z-drugs", such as zopiclone and zolpidem, which together it calls benzodiazepine receptor agonists. It applies to people taking them for insomnia, either on its own or alongside other conditions that are well managed. It does not apply to people with other sleep disorders, or with untreated anxiety, depression or other conditions that may be causing or worsening insomnia. ^[1] Other types of sleep aid are not covered here.

The American Academy of Sleep Medicine (AASM) guideline on medicines for chronic insomnia (2017) reviewed zolpidem, eszopiclone, zaleplon, temazepam, triazolam and several other drugs, but did not assess zopiclone. It rates every one of its medicine-specific recommendations as "weak", which it says reflects limited, low-quality evidence rather than proof that a medicine does not work. ^[11]

What the research says

The research is about supervised reduction. Canada's 2018 deprescribing guideline recommends offering a slow taper to adults 65 and over, and suggests it for adults aged 18 to 64 after more than four weeks of use; its trial review found tapering helped more people stop than usual care, without serious harms. ^[1] In one trial, 76 older adults averaging 19 years on benzodiazepines had a supervised taper, CBT-I, or both; after ten weeks 85% of the combined group were off the medicine, against 48% and 54% for taper or CBT-I alone, and the gains held at 12 months. ^[2] A trial of 53 long-term users found sleep improved more when self-help CBT accompanied the taper. ^[5] Among 303 older long-term users, patient education alone led 27% to stop by six months, against 5% of controls. ^[4] One caution: the guideline found CBT's extra boost did not last long term. ^[1] In plain terms, a prescriber-led taper helps more people stop; nothing here is a reason to change a medicine yourself.

The studies behind this

• Canadian guideline (Pottie, 2018): developed with the GRADE method from a systematic review of trials. It recommends offering a slow taper to adults aged 65 and over, whatever their length of use, and suggests offering it to adults aged 18 to 64 who have taken these medicines for more than four weeks. It found that tapering improved the number of people who stopped, compared with usual care, without serious harms. ^[1]
• Morin trial (2004): 76 older adults who had used benzodiazepines for sleep for an average of 19 years were randomly assigned to a supervised taper, to CBT-I, or to both. After the 10-week program, 85% of the combined group were benzodiazepine-free, compared with 48% of the taper-only group and 54% of the CBT-I-only group. The early reductions in drug use were well maintained at 12 months, and sleep improvements became more noticeable over that time. In this supervised setting, no significant withdrawal symptoms or adverse events were linked to tapering. ^[2]
• Belleville trial (2007): 53 chronic hypnotic users followed an eight-week taper program, either alone or with self-help CBT. Medicine use fell in both groups, and sleep improved more in the group that also had CBT. ^[5]
• EMPOWER trial (Tannenbaum, 2014): 303 long-term benzodiazepine users aged 65 to 95, recruited through 30 community pharmacies. A patient education program describing the risks and a stepwise taper led 27% of the intervention group to stop by six months, against 5% of the comparison group. 62% of the intervention group started a conversation with a doctor or pharmacist about stopping. ^[4]
• A caution from the guideline: its review found that adding CBT improved the number of people who stopped compared with tapering alone, but that the improvement was not maintained in the long term. ^[1]

Why do people want to stop?

• Benefit may fade while side effects do not. The Canadian guideline notes that guidelines suggest these medicines be used short term (typically up to four weeks) and that evidence suggests their effect on insomnia can diminish within about four weeks, while adverse effects may persist. ^[1] The AASM guideline makes a related point from the trial literature: the gains from CBT-I tend to last after treatment ends, whereas those from medication tend to fade once the drug is stopped. ^[11]
• Harms listed in the Canadian guideline: physical dependence, drowsiness, balance problems, falls, fractures, cognitive impairment, memory problems and motor vehicle accidents. Much of this evidence is observational and comes mainly from older adults. ^[1] The AASM guideline says the risks of dependency and serious withdrawal complications are of greatest concern with true benzodiazepines, particularly with escalating long-term use and insufficient monitoring. ^[11]
• Small average benefit in older adults: a review of 24 trials (2,417 people aged 60 and over) found sleeping pills added about 25 minutes of sleep compared with placebo. Adverse cognitive events were 4.78 times as common, and daytime fatigue 3.82 times as common. The authors concluded that the benefits may not justify the risks in people over 60. ^[6]
• Not everyone wants to stop. The guideline notes that some people would like to stop but worry about insomnia returning, and others state strongly that they want to continue. ^[1]

What can happen when people stop?

• Rebound insomnia. Rebound means symptoms return after stopping, at greater severity than before treatment. ^[8] A meta-analysis of sleep laboratory studies of short-acting hypnotics found that rebound on the first night after stopping was intense with triazolam and mild with zolpidem. Data were inadequate for zopiclone. ^[7] The AASM guideline's own reading of the zolpidem trials is mixed: some found no rebound after up to six months of use, while others found rebound limited mainly to the first night. ^[11]
• Withdrawal symptoms. The Canadian guideline lists irritability, sweating, stomach and bowel symptoms and anxiety. It describes them, when they occur, as typically mild and short-term, lasting days to weeks. It also found that tapering might cause more sleep problems than carrying on, with no difference between the groups after 12 months. ^[1]
• Serious withdrawal. A 2025 review explains that stopping suddenly can trigger withdrawal syndromes, and that classic withdrawal syndromes can include seizures and delirium. ^[8] The Canadian guideline says seizures do not appear to occur with tapering but have been reported rarely in people who stopped very high doses without tapering, or who had an underlying seizure disorder. ^[1]

Because stopping suddenly can trigger serious withdrawal, the research is about supervised reduction, and not about stopping suddenly.

Where does CBT-I come in?

The AASM behavioural guideline notes that CBT-I may reduce the need for medication and therefore reduce the risk of drug-related harm. ^[9] Its companion guideline on medicines says that CBT-I should be the initial treatment for chronic insomnia, with medication considered mainly for people who cannot take part in CBT-I, who still have symptoms despite it, or in selected cases as a temporary addition to it. ^[11] In the Morin trial the groups that received CBT-I reported larger sleep improvements than taper alone. ^[2]

A Norwegian trial in which 125 people who had been prescribed zopiclone or zolpidem completed follow-up compared a sleep hygiene sheet with a self-help book based on CBT-I. The share taking medicine daily fell in the book group, from 25.4% to 18.6%. The lead author wrote the book, and outcomes were self-reported, so treat it as supporting evidence only. ^[10]

Staying off: how durable is it?

Not always. In a two-year follow-up of 47 older adults who had successfully stopped, 42.6% had restarted benzodiazepines. Relapse was lower in the groups that had a supervised taper, with or without CBT-I (about a third), than in the CBT-I-only group (about two-thirds). The authors suggest booster sessions might help. ^[3]

For clinicians

The Canadian guideline (Pottie and colleagues, 2018) is the evidence-based starting point. It sets out who it applies to, when deprescribing should be offered, the harms of continued use, and examples of tapering and monitoring. We deliberately do not reproduce taper schedules here. ^[1] The AASM pharmacologic guideline (2017) is the American counterpart for the medicines themselves. ^[11] Small Hours has a page for clinicians whose patients use the program.

Where the evidence is thin

• Many of the key trials were small and recruited older, very long-term users, so results may not fit everyone. ^[2][3][5]
• Trials used different taper protocols, the main supervised-taper trials here involved benzodiazepines, and the rebound data were inadequate for zopiclone. ^{[1][2][3][4][7]}
• Findings on whether CBT-I adds lasting benefit to a taper are mixed. ^[1][2][3]
• Much of the evidence on harms of continued use is observational. ^[1]
• The AASM medicines guideline rates all of its drug-specific recommendations as weak, and it does not cover zopiclone. ^[11]
• Relapse after stopping is common. ^[3]

Where does this leave you?

If you have been taking a sleeping pill for a long time and wonder about it, the research supports starting that conversation with your doctor or pharmacist, rather than acting alone. Small Hours does not give medication advice. It is a six-week CBT-I program that works from your own sleep diary, which you can use alongside the care you already have. You can sign up at trysmallhours.com.

Again: do not stop or change a sleeping medicine without your doctor or pharmacist, and remember that stopping some medicines suddenly can be dangerous.

General education, not medical advice.