Safety
Zopiclone, zolpidem, diazepam. The short-term role, the long-term problem, and how to stop safely.
LoveMyLife Sleep Medicine team
MRCGP-led, respiratory consultant-overseen
22 April 2026
6 min read

Z-drugs (zopiclone, zolpidem) and benzodiazepines (diazepam, temazepam, lorazepam, nitrazepam) are the oldest-established sleep medications still widely prescribed in the UK. They work. They also have a dependence problem that is under-recognised by both prescribers and patients. This article covers the short-term role, the long-term harms, and how to come off safely if you have been on them a long time.
Both drug classes act on GABA-A receptors. GABA is the brain's main inhibitory neurotransmitter; potentiating its action produces sedation, anxiolysis, and muscle relaxation. Z-drugs are more selective for the sleep-relevant GABA-A subtypes; benzodiazepines hit a broader range.
Functionally, both produce sleep. The subjective experience is different: Z-drugs tend to produce sleep that feels closer to normal; benzodiazepines produce sleep that can feel more drugged and more next-day sedating.
Z-drugs were introduced in the 1990s as "non-benzodiazepines" with a better safety profile than benzos. The marketing emphasised lower dependence risk and cleaner next-day function. The reality is that Z-drugs are benzodiazepines with a slightly different shape. The dependence risk is lower but not low. The long-term effects are similar.
In real-world NHS prescribing, Z-drug use often extends for years, and patients become physically dependent. The 4-week NICE recommendation is routinely overrun.
For acute insomnia with a clear short-term cause (bereavement, major surgery, travel-related disruption, short-term work crisis), 2 to 4 weeks of Z-drug use can prevent acute insomnia becoming chronic. The evidence for this is reasonable and the risk of dependence at that duration is low.
For in-hospital use (peri-operative, ICU step-down), short courses are standard and appropriate.
For patients with rapidly-resolving medical causes of insomnia (pain that will resolve in days, acute illness), short-course use is appropriate.
For jet lag with severe disruption (rare), 1 to 2 nights use may be reasonable.
Continuous nightly Z-drug use beyond 4 to 6 weeks produces:
- Tolerance: the same dose becomes less effective; patients often escalate dose to maintain effect - Physical dependence: missed doses produce rebound insomnia, anxiety, and sometimes withdrawal symptoms - Cognitive impairment: memory, attention, and processing speed are measurably impaired by long-term Z-drug use, particularly in older adults - Fall risk: significantly increased, particularly in older adults - Association with dementia risk: observational evidence suggests long-term hypnotic use is associated with increased dementia incidence, though causality is debated - Daytime mood effects: apathy, emotional blunting in some patients - Parasomnia risk: sleep walking, sleep eating, sleep driving - rare but documented
Benzodiazepines carry all these with additional withdrawal complexity (more prolonged and sometimes more severe).
The medication works. Acutely, it still produces sleep. Stopping produces immediately worse sleep.
The prescription is easy. A 28-day script is faster than addressing the underlying cause.
The rebound on stopping is unpleasant. The first 1 to 2 weeks after stopping are typically worse than usual sleep.
The alternatives take work. CBT-I, lifestyle change, and treating underlying causes are slower and more demanding than taking a tablet.
Medical inertia. GP caseloads are full; continuing an existing prescription is the path of least resistance.
You are probably dependent if:
- You have been using the medication nightly for 3 months or more - Missing a dose produces noticeable sleep problems - You feel anxious about not having the medication when travelling - The dose has crept up over time - You have tried to stop and found it unmanageable - You have run out and noticed physical symptoms (tremor, anxiety, sweating) beyond just poor sleep
Dependence is not moral failure. It is a predictable pharmacological response to long-term use.
Slow taper. Abrupt cessation is usually unnecessary and counterproductive. Typical taper: reduce dose by 10 to 25 percent every 2 to 4 weeks. For zopiclone 7.5mg, drop to 3.75mg, then 1.875mg, then alternate nights, then stop. Timescale: 2 to 4 months for most users.
Concurrent CBT-I. Rebuilding the behavioural substrate so that sleep can happen without medication is essential. Taper without behavioural work usually fails. CBT-I started at the beginning of the taper and maintained throughout is the evidence-based approach.
Support through the first difficult weeks. The first 2 to 4 weeks of taper typically include worse sleep, increased anxiety, and sometimes brief exacerbations. Planning for this, having unlimited clinician messaging support, and having the option to slow the taper if it becomes unmanageable all help.
For benzodiazepines, longer timescales. Diazepam and other long-half-life benzodiazepines need tapers over 6 to 12 months, often with dose conversion to a long-acting agent first (e.g. temazepam to diazepam). This is specialist work; not every clinician is comfortable running it.
For patients on long-term Z-drugs or benzodiazepines, we offer a structured coming-off programme combining medication taper with CBT-I and medical oversight. Typical timeline: 12 to 16 weeks. Included in our ongoing care plan (£95 per month). Weekly messaging support, fortnightly clinician reviews, emergency response if taper is not going well.
For benzodiazepine dependence specifically, we may work with specialist pharmacy colleagues or refer to NHS substance-misuse services if the dependence is severe or complex.
We do not repeat-prescribe Z-drugs long-term. If you come to us for ongoing Z-drug prescriptions as the primary request, we will ask what you are trying to achieve and offer the coming-off programme instead.
A small minority of patients genuinely benefit from continuing low-dose Z-drug use long-term. Typically: older adults with specific insomnia phenotypes that do not respond to CBT-I, patients with severe treatment-resistant insomnia where the alternative is worse, patients with specific neurological conditions. These decisions are individual and usually specialist.
CBT-I. First-line, see dedicated article.
Melatonin (prolonged release). Some effect in older adults, minimal in working-age.
Trazodone or mirtazapine. Useful where there is coexisting depression.
Treating the underlying cause. Often apnoea, menopause, or depression is the actual problem that looked like insomnia.
Z-drugs and benzodiazepines have a legitimate short-term role. Long-term use produces dependence with real costs. Most patients who have been on them for years can come off safely with structured support and concurrent CBT-I. If you are on long-term Z-drugs, a conversation about alternatives is worth having.
Clinically reviewed
Dr Seth Rankin · MBChB MRCGP - Founder and Medical Director, LoveMyLife
5 services
If this article has made you think our assessment might help, the next step is a short consultation with one of our sleep-medicine doctors.
Begin your consultation at this link. Online with a WatchPAT One home study, or in person at Westfield London.