Stopping sleeping pills: what Z-drugs achieve, what risks they carry and what a supervised way out can look like

It is just after eleven. You are lying in bed, and your mind is still running. On the bedside table lies the small packet.

You know that with a pill you will soon be asleep. And that this is exactly what you did not want anymore.

Many people know this pattern. The pill was meant for a few nights, after a death, a breakup, an operation. Then a few nights became a month. And the month became a year.

If you recognise yourself in this: you have done nothing wrong. There were usually good reasons for the prescription. Sleeping pills can be a real bridge in an acute crisis. And the fact that the body adapts to a substance is not a question of character, it is pharmacology.

At the same time, you deserve honest numbers: what these drugs achieve in the sleep lab, which risks are documented and what studies show about a way out that is not abrupt, but supervised.

How Z-drugs and benzodiazepines work: one switch, many names

Perhaps you have heard that zolpidem or zopiclone are something different from the old tranquillisers. More modern, more targeted, more harmless. Whether the pharmacology supports that becomes clear when you look at the switch both groups act on.

Your brain works with an accelerator and a brake. The most important braking signal is called GABA. Benzodiazepines and Z-drugs do not press the brake themselves. They act more like a brake booster: the body's own GABA gets more force.

Half-life: why the morning and the night after stopping depend on it

A long time in the body means more residual effect in the morning, and the S3 guideline names hangover effects especially with long half-lives. A short time in the body means the brake can fall away more quickly, and according to the prescribing information for zopiclone, temporary discontinuation symptoms can then occur even after shorter treatment, above all with short-acting benzodiazepines.

And now you know why two drugs with completely different names can cause such similar problems.

What sleeping pills achieve, measured in minutes

The first night with a pill often feels like a release. That feeling is real. Still, it is worth asking how much of it a sleep lab measures.

The 22 and the 4.2 minutes come from different studies with different methods. They stand side by side here, not against each other. In a study cited in the prescribing information for zolpidem, involving 462 healthy volunteers with transient insomnia, the time to fall asleep shortened by 10 or 3 minutes, depending on the dose.

Why the benefit feels larger than it measures

If you are thinking, but for me it is much more than twenty minutes: there are at least three explanations. First, the placebo response. Alongside the drug effect, it is a component of the improvement you experience in its own right, a real reaction to expectation and ritual. Second, the way it is measured. Estimated and measured time to fall asleep often diverge, and how large such gaps can be is something I describe in the article Sleep trackers: what they measure and how accurate they are. Third, perception itself.

And now you know why a pill that brings only a few minutes in the lab can feel like the difference between night and day in bed.

Tolerance, dependence and why the limit is four weeks

At first it goes quickly. Pill, half an hour, sleep. After a few weeks you are lying awake again despite the pill. Or you notice already in the afternoon that the evening without it worries you. Both have been in the prescribing information for years.

Tolerance: the same pill, less sleep

According to the prescribing information for zolpidem, repeated use of short-acting benzodiazepines or benzodiazepine-like substances over a few weeks can lead to a loss of effectiveness. For zopiclone, the prescribing information states that no notable tolerance developed during four weeks of treatment. In sleep laboratory studies, marked tolerance developed with triazolam, but only slight tolerance with zolpidem.

Dependence: the body has adjusted

Dependence here means, first of all, something physical. The nervous system can get used to the boosted brake and come to count on it. The prescribing information for zolpidem states that taking benzodiazepines can lead to physical dependence even at therapeutic doses and that the risk rises with dose and duration. The body may have adapted after just a few weeks of regular use. The data do not support a fixed number of weeks for this.

The four-week limit is no coincidence. Both sets of prescribing information limit treatment to four weeks, including the gradual withdrawal phase, and according to the zopiclone prescribing information patients should be informed about this right at the start. According to both, longer treatment may be necessary in certain cases, but not without a renewed medical assessment. The S3 guideline names one exception: for eszopiclone, the prescribing information notes that in certain cases it can be taken for up to a maximum of six months.

And now you know why the prescribing information says four weeks from the start, and why stopping is already counted in that calculation.

The risks in numbers: falls, accidents, night-time actions, memory

At night, on the way to the bathroom. The hallway is dark, your legs feel weak, your head feels wrapped in cotton wool. A moment of unsteadiness is enough.

First of all: most of the risk data come from observational studies. They show associations, not proven causes, because people taking sleeping pills are often older, sicker or under more strain than people who are not.

Falls and fractures

Two further studies fit with this. A meta-analysis of 14 observational studies led by Treves found an increased risk of fractures with Z-drugs (odds ratio 1.63 in 830,877 people), while the estimate for falls was 2.40 but not significant. And an English cohort study led by Richardson in 27,090 people with dementia found more fractures, hip fractures, falls and ischaemic strokes at higher doses (risk estimates 1.67, 1.96, 1.33 and 1.88), and only small or inconsistent excess risks at lower doses. Compared with benzodiazepines, no difference in adverse effects was found there, except for lower mortality with Z-drugs.

For older people, the S3 guideline names falls, confusion, delirium and cognitive impairment as side effects of almost all sedatives and recommends medication in older people only with great restraint. More on sleep in later life in the article Sleep disorders in older age.

Road traffic and the morning after

In a cohort led by Hansen of 409,171 adults, 5.8 percent of whom received a new prescription for a sleep medication, the hazard ratio for road traffic crashes with new use compared with non-use was 1.27 for temazepam, which was not significant, 1.91 for trazodone and 2.20 for zolpidem. The authors equate these values with blood alcohol concentrations between 0.06 and 0.11 percent. For zolpidem, the prescribing information states an increased risk of impaired driving ability if less than 8 hours pass between taking it and activities that require heightened attention. The zopiclone prescribing information names at least 12 hours between taking it and driving. Whether you can drive should be clarified with a doctor.

Sleepwalking, sleep eating, sleep driving

The prescribing information for zolpidem describes driving while sleepwalking, referred to there as sleep driving, as well as preparing and eating meals, making phone calls or having sex without any later memory, and this after the first or any subsequent dose.

In Germany, too, previous unusual sleep behaviour after taking the drug is a contraindication according to the prescribing information for zolpidem and zopiclone. If you notice something like this in yourself or in the person next to you, it should be discussed with a doctor promptly. According to the prescribing information for zolpidem, stopping should then be strongly considered, and that decision is made by the treating doctor together with you.

Memory and thinking

This fits with the 4.78 times more frequent cognitive side effects in the meta-analysis by Glass and with the memory gaps for the period after taking the drug, which the prescribing information for zopiclone describes under the heading amnesia.

And now you know why guidelines are so cautious, especially for older people, even if a pill can make a single night easier in individual cases.

Do sleeping pills cause dementia? What observational data show and what they do not

Perhaps you have read a headline linking sleeping pills to dementia. Since then, a second worry comes with every pill. As things stand today, the answer is: the question is disputed, and all studies on it are observational data.

Why these data are so hard to read

Imagine an old house in which a cable has been smouldering for years. Long before anyone finds the damage, the lights flicker. Insomnia, anxiety and low mood can be such early flickering signs of beginning dementia. Anyone who is given a sleeping pill because of them later appears in the statistics as a user with dementia, even if the pill only accompanied the flickering. Experts call this reverse causation.

For older people, the S3 guideline names cognitive impairment up to drug-induced dementia as a side effect of almost all sedatives. This means medication-related impairment of thinking and memory, not Alzheimer's disease.

And now you know why a serious answer to the dementia question should neither reassure nor alarm.

What the guidelines say, and why they are so cautious

Perhaps you have experienced one practice prescribing a sleeping pill without hesitation and the next being very reluctant. Both can have good reasons. The guidelines show the professional framework.

The recommendation against long-term treatment received majority agreement. Under the AWMF rules the guideline refers to, this is the lowest level of consensus, with more than 50 percent of votes, while consensus means more than 75 and strong consensus more than 95 percent. On combining a pill with CBT-I, the guideline cites two studies that suggest a synergistic effect in the acute phase, and afterwards CBT-I alone was superior to the combination.

Why this caution? Not out of principle against medication. According to the guideline, sleep problems return in many cases after drug treatment is stopped, which creates the risk of dependence, and the evidence on long-term use is insufficient for all approved substances.

The European guideline led by Riemann is less strict here than the German one. The AASM guideline led by Sateia emphasises that a weak recommendation expresses lower certainty, not ineffectiveness.

I will not describe the first choice of the German and European guidelines, CBT-I, again here. How sleep restriction, stimulus control and work on night-time rumination loops fit together is covered in the article CBT-I and sleep restriction.

And now you know why the same pill can be a justifiable bridge in one situation and, in another, a question that deserves a fresh conversation.

Stopping sleeping pills: why never abruptly, what rebound insomnia is and what studies on tapering show

Perhaps you have tried it before. One evening without a pill. And then that night: wide awake at two, heart pounding, thoughts going round in circles. At some point the sentence in your head: it just does not work without. The next evening the pill was back on the bedside table.

If this sounds familiar: that night proves less than it claims. And abruptly leaving out the pill after longer use is exactly the route that the prescribing information and guidelines do not intend. Please do not repeat such an attempt on your own, but bring it up at your next appointment.

Why the body protests when the brake suddenly disappears

Imagine a tug of war. The sleeping pill pulls on the side of calm, and your nervous system has pulled against it for weeks. If the pill suddenly lets go, the counterweight can stumble backwards.

When treatment is stopped suddenly after physical dependence, the prescribing information for zolpidem states that in severe cases a sense of losing touch with reality, hypersensitivity to light and noise, hallucinations, delirium or epileptic seizures can occur. The prescribing information for zopiclone also names hallucinations and epileptic seizures and, because the risk is higher especially after longer treatment, recommends reducing the dose gradually.

Rebound insomnia: why the first night can be misleading

According to the prescribing information for zolpidem, temporary discontinuation symptoms can occur when treatment ends, and the complaints that led to treatment can return in a stronger form. This is rebound insomnia.

Why does this particular night seem so convincing? Because it feels like confirmation. Yet part of that night can be the rebound itself, a temporary countermovement and not a statement about how you might sleep in a few months. The prescribing information for zolpidem explicitly states that patients should know about rebound phenomena in order to minimise anxiety about the symptoms.

I have found no robust figures on how long rebound lasts after long-term use. Two of the following studies do suggest, however, that sleep can continue to improve over months after a supervised way out.

What studies on supervised tapering show

The following results are literature, not instructions. I deliberately leave out how the programmes were structured.

A meta-analysis led by Parr on discontinuation strategies in general practice and outpatient care found gradual dose reduction clearly superior to routine care (odds ratio 5.96), as were brief interventions (4.37). An abrupt substitution of the benzodiazepine with another medication, by contrast, performed worse than gradual reduction alone (0.30). In a Dutch general practice study of 180 long-term users who wanted to stop, discontinuation succeeded in 62 percent with structured tapering and in 21 percent with usual care.

Why cognitive behavioural therapy belongs in the process of stopping

Taking a pill away is only half the task. The other half is what takes its place at night.

In the meta-analysis by Parr, psychological treatment plus gradual reduction was superior to reduction alone (odds ratio 1.82). A Cochrane review led by Darker of 25 randomised trials found the advantage of CBT plus tapering in the short term (risk ratio 1.40 up to four weeks after treatment, 1.51 after three months), but no longer after six months. The 2025 tapering guideline recommends offering behavioural interventions alongside tapering, explicitly including CBT-I. How CBT-I is structured is explained in the article CBT-I and sleep restriction for chronic insomnia.

What is not proven as an aid to stopping

Magnesium for sleep problems in general is a different question, which I discuss in the article Magnesium for sleep disorders. As an aid to stopping, it is not proven.

Melatonin as an aid to stopping: mixed. In an early small study led by Garfinkel, 14 of 18 people taking prolonged-release melatonin stopped the benzodiazepine, but only 4 of 16 on placebo. In a study led by Vissers in Dutch general practices, by contrast, 40 percent in both groups had stopped after one year, in 86 people with schizophrenia or bipolar disorder in a study led by Baandrup the discontinuation rate was 38.1 percent with melatonin and 47.7 percent with placebo, and in the study led by Lähteenmäki melatonin brought no advantage either. An expert panel led by Palagini takes a more optimistic view in 2025. Any decision about an additional active substance belongs in a doctor's hands.

Clinical tradition without a strong study base: I have found no verified study on GABA capsules or L-theanine as an aid to stopping. Their benefit is open, and they do not replace medically supervised tapering.

And now you know why getting off sleeping pills is less a question of willpower than a question of pace, support and what takes the place of the pill at night.

Alternatives to sleeping pills: what is proven and what is not

If not the pill, then what? First of all: none of the following options is a guide to switching. Changing to another medication is a medical decision.

CBT-I: the first choice

Supported by meta-analyses and guidelines: the S3 guideline recommends CBT-I to all people with insomnia as the first treatment option, preferably in person, and the European guideline in person or digitally. How it works is explained in the article CBT-I and sleep restriction.

Sleep-promoting antidepressants

The S3 guideline rates doxepin and trazodone as effective in off-label treatment, but advises against long-term treatment. According to the guideline, they are not approved in Germany for insomnia without an accompanying depressive disorder. And in the traffic data from Hansen, trazodone, which is often considered a more harmless alternative, was also linked to more crashes.

Prolonged-release melatonin from age 55

According to the S3 guideline, melatonin is approved in Germany for the short-term treatment of insomnia from age 55, but the guideline does not recommend long-term treatment. The European guideline names prolonged-release melatonin for up to three months from age 55, and the network meta-analysis by De Crescenzo found no meaningful overall benefit for melatonin. More in the article Melatonin: effects, dosage and myths.

Orexin receptor antagonists: a newer class of drugs

Orexin is a messenger that maintains wakefulness. Orexin receptor antagonists block its receptors, so they ease off the accelerator instead of boosting the brake. From this class, the S3 guideline names daridorexant, which according to the guideline was approved by the EMA in 2022.

In an extension study with manufacturer involvement led by Kunz, involving 804 people over 40 weeks, neither withdrawal symptoms nor rebound occurred in the subsequent placebo phase, which lasted only seven days. Two network meta-analyses contradict each other: Yue saw the class ahead of the Z-drugs for wake time and sleep efficiency, while De Crescenzo counted daridorexant among the drugs with poor tolerability or missing long-term data. The S3 guideline rates the class as effective but does not recommend long-term treatment, especially not beyond one year. It also criticises that subjectively experienced difficulties falling and staying asleep were not among the primary or secondary endpoints in the approval studies. So this class, too, is not a sleeping pill without open questions.

Herbal remedies and what is not recommended

Guideline position: according to the S3 guideline, treating insomnia with herbal medicinal products should not be recommended. Valerian showed at most a slight superiority over placebo, with low study quality. The European guideline also does not recommend phytotherapeutics, nor antihistamines, antipsychotics, fast-release melatonin or ramelteon. More on this in the articles Natural and herbal sleep aids and Valerian and passionflower.

And now you know why the search for the perfect replacement pill often starts in the wrong place.

Looking behind the insomnia and preparing the conversation

A sleeping pill answers one question: how do I get some rest tonight? Another question often gets left behind: why did my sleep tip over in the first place?

This second question does not replace tapering or CBT-I, but it belongs alongside them. According to the S3 guideline, CBT-I is the first option even for insomnia with coexisting conditions, and those coexisting conditions deserve their own treatment.

Causes a pill can mask

• Sleep apnoea. Pauses in breathing at night can chop sleep into pieces, and sedating drugs are a delicate matter here: the prescribing information for zopiclone lists severe sleep apnoea syndrome as a contraindication, and the one for zolpidem lists sleep apnoea syndrome. More in the article Recognising sleep apnoea.
• Iron deficiency and thyroid. Both can favour waking at night, see Iron deficiency and thyroid as sleep thieves.
• Depression. Insomnia can be a sign and an amplifier, and depression needs medical or psychotherapeutic treatment, see Sleep and depression.
• Menopause. Hormonal change can alter sleep considerably, see Menopause and sleep disorders.
• Stress and waking at night. See Awake at 3 a.m., and for ongoing exhaustion Burnout and sleep disorders.
• Falling asleep or staying asleep. See Trouble falling asleep and Trouble staying asleep, and for the basics Putting sleep hygiene into practice.

The complete map of causes is in the overview article Treating sleep disorders holistically.

Questions for your conversation

In the EMPOWER trial, well-informed people sought a conversation about their sleeping pills much more often. The following questions are meant as a guide for the conversation, not as instructions.

And now you know why the way out of the pill does not lead through one brave evening, but through a well-prepared conversation.

Frequently asked questions about sleeping pills and stopping them

Where the topic of sleeping pills connects with other questions

Sleeping pills are rarely the whole story. Behind them lie questions about causes, stages of life and the nervous system. These articles take things further from here.