Temazepam: Independent Evidence on Insomnia, Side Effects & Withdrawal

Table of contents

• Evidence summary
• What temazepam is
• How it works
• What it is prescribed for
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions
• Who should avoid temazepam
• Dosage and how to take it
• Follow the money: who makes it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

Independent evidence and credibility scorecard

Independence tier: 1 = government regulator or publicly funded with no declared industry ties to the drug; 2 = public or academic funding, but some authors declare industry ties (mostly unrelated to temazepam); 3 = professional society that also receives industry money; 4 = manufacturer-authored or manufacturer-submitted. Credibility (A–D) combines independence with study design and relevance to temazepam.

What temazepam is

Temazepam is a benzodiazepine hypnotic (sleeping medicine). Its ATC code is N05CD07, and the UK SmPC says its actions are similar to oxazepam and diazepam: "central nervous system sedation, anxiolysis and muscle relaxation" (UK SmPC). The US brand is Restoril, sold as capsules of 7.5 mg, 15 mg, 22.5 mg and 30 mg (FDA label).

• US approval: the FDA first approved Restoril (NDA 018163) on 27 February 1981, as a new molecular entity (Drugs@FDA). The Drugs@FDA record does not name the original 1981 applicant, and we could not verify it from a primary source. A 2002 FDA letter to Mallinckrodt records the deletion of "SandoPak" packaging information from the label (FDA letter, 2002). This suggests, but does not prove, an earlier Sandoz link.
• Current US holder: SpecGx LLC (Webster Groves, Missouri, US). The 2026 label is "© 2026 Par Health, Inc." and says "Product of Italy" (FDA label).
• Generic status: temazepam has been generic for decades. Drugs@FDA lists Restoril capsules with an "AB" therapeutic-equivalence code, which means approved generics are rated equivalent (Drugs@FDA). In the UK, the emc database currently lists one temazepam product with a full SmPC: an oral solution (10 mg/5 ml) from Syri Ltd, trading as Thame Laboratories / SyriMed, Ruislip, UK (UK SmPC).
• Legal status, UK: prescription-only and listed in Schedule 3 of the Misuse of Drugs Regulations 2001 (legislation.gov.uk). It is also one of the drugs with a specified blood limit under drug-driving law: 1,000 µg/L (Drug Driving Regulations 2014).
• Legal status, US: prescription-only and a Schedule IV controlled substance (FDA label).

How it works

Like other benzodiazepines, temazepam enhances the brain's main calming chemical messenger, GABA. The FDA describes the class as working by "binding to gamma-aminobutyric acid (GABA) receptors in the brain" to slow brain activity, "which causes drowsiness or calming effects" (FDA 2020). Opioids act mainly at mu receptors, so the two drug types depress breathing through different receptor systems. According to the US label, this is why combining them can "significantly worsen opioid-related respiratory depression" (FDA label).

What the body does with it

• Absorption: after a 30 mg capsule, the drug is measurable in blood within 10–20 minutes. Peak levels come at about 1.2–1.6 hours (mean 1.5 hours). First-pass metabolism is minimal (8%) (FDA label).
• Half-life: the US label gives a terminal half-life of 3.5 to 18.4 hours (mean 8.8 hours). The UK SmPC reports 5.3–11.5 hours after night-time dosing in young volunteers. In older people it reports a mean of about 15 hours, possibly longer in older women (UK SmPC).
• Metabolism: temazepam is "completely metabolized through conjugation" before excretion. It has no active metabolites, and 80–90% of the dose appears in urine (FDA label). This is why it has fewer liver-enzyme interactions than some benzodiazepines, although the UK SmPC still lists CYP450 inhibitors and inducers as a consideration.
• Sleep architecture: in the US label's sleep-laboratory studies, REM sleep was "essentially unchanged" and slow-wave (deep) sleep "was decreased" (FDA label).

The US label also explains a problem with short-to-intermediate-acting hypnotics. After several weeks of nightly use, the drug level can dip between doses. The label links this to "increased wakefulness during the last third of the night, and the appearance of increased signs of daytime anxiety" (FDA label). The UK SmPC likewise warns that with short-acting benzodiazepines such as temazepam, withdrawal "can become manifest between doses, especially when the dosage is high" (UK SmPC).

What it is prescribed for

Licensed uses

• UK: "short term treatment of sleep disturbances, considered severe or disabling or where insomnia is subjecting the individual to extreme distress". It is also licensed as pre-medication before minor surgical and investigative procedures (UK SmPC).
• US: "short-term treatment of insomnia (generally 7 to 10 days)". The trials supporting approval "were 2 weeks in duration" (FDA label).

Where guidelines place it

What works and what does not

Benefits by claim

Short-term sleep onset and duration

The most direct independent synthesis of temazepam data is the AASM guideline's review. It found three RCTs of temazepam in chronic insomnia, with only 19, 20 and 34 participants. For 15 mg, the pooled self-reported reduction in time to fall asleep was 20.06 minutes (CI −1.07 to −39.05), and the pooled self-reported increase in total sleep was 64.4 minutes (CI +8.1 to +120.8). The task force judged both to exceed its threshold for clinical significance. The confidence intervals are wide, though: at one end, the time-to-sleep improvement is barely one minute. One trial (Wu) reported a polysomnography-measured increase in total sleep of 99.1 minutes versus placebo at 15 mg (AASM 2017).

The placebo response is large in insomnia trials. In one temazepam 20 mg sleep-laboratory study (Tuk) of "primary sleep onset insomnia", both temazepam and placebo shortened sleep latency by about 53 minutes from baseline, with no difference between them (AASM 2017). The manufacturer's own label, describing the 2-week lab studies that supported approval, reports a similar pattern. Significant drug–placebo differences at 2 weeks occurred "only for total sleep time at the 2 higher doses, and for sleep latency only at the highest dose" (FDA label).

Class-level meta-analyses point the same way. Holbrook and colleagues pooled 45 benzodiazepine RCTs (2,672 patients). In sleep-record data, benzodiazepines increased total sleep by 61.8 minutes (95% CI 37.4 to 86.2), but the 4.2-minute reduction in time to fall asleep was not statistically significant. Patients' own estimate was a 14.3-minute reduction. Daytime drowsiness and dizziness were more common (common OR 1.8) (Holbrook et al., CMAJ 2000). The 2022 Lancet network meta-analysis confirmed that benzodiazepines beat placebo for acute treatment. It also found that benzodiazepines, eszopiclone, zolpidem and zopiclone produced more people with side-effects than placebo (OR range 1.27–2.78) (De Crescenzo 2022).

Sleep quality

For 15 mg, the AASM meta-analysis found a sleep-quality improvement of 0.25 standard deviations. The task force said this "falls below the threshold for clinical significance". It noted that one of the two trials was underpowered (AASM 2017). In older adults on any sedative hypnotic, Glass et al. found an effect size of 0.14 for sleep quality (Glass 2005).

Is the benefit worth it?

The most useful single figure for patients comes from the unfunded Glass meta-analysis in people aged 60 and over. 13 people would need to take a sedative hypnotic for one to report better sleep quality, while one in every 6 would have an adverse event. The authors concluded that "an adverse event is more than twice as likely as enhanced quality of sleep" and that in people over 60 "the benefits of these drugs may not justify the increased risk" (Glass et al., BMJ 2005). That analysis pooled several drugs, and temazepam was one of them. For younger adults the harm profile is milder, but the regulatory ceiling of a few weeks still applies.

Compared with CBT-I

In an NIMH-funded trial of 78 older adults, all three active arms (CBT, temazepam, and the two combined) beat placebo after 8 weeks. The reduction in time awake after sleep onset was 63.5% for the combination, 55% for CBT, 46.5% for temazepam and 16.9% for placebo. CBT gains were sustained over up to 24 months of follow-up, while temazepam-alone gains were not. Participants, their partners and clinicians all rated the behavioural treatment as more effective (Morin et al., JAMA 1999). This is why guidelines place CBT-I first and reserve drugs like temazepam for short, bridging courses.

Risks and all side effects

FDA boxed warning (summary of the label wording)

The FDA imposed this class-wide update in September 2020. It said the previous labelling "does not provide adequate warnings". It also noted that physical dependence "can occur when benzodiazepines are taken steadily for several days to weeks, even as prescribed". Its data review found about 92 million US benzodiazepine prescriptions in 2019. About 50% of patients dispensed oral benzodiazepines in 2018 received them for two months or longer (FDA 2020). Drugs@FDA records the resulting Restoril label update approved on 5 February 2021 (FDA approval letter S-067).

UK regulator (MHRA)

In March 2020 the MHRA reminded prescribers that benzodiazepines and opioids "can both cause respiratory depression" and that the effects are additive. The reminder followed a coroner's report of a death from respiratory arrest in a man given clonazepam and methadone. The advice is to co-prescribe only if there is no alternative, to use the lowest doses for the shortest time, and, with methadone, to monitor for at least 2 weeks after starting or changing treatment (MHRA Drug Safety Update).

Common side effects (US controlled trials)

In controlled studies, 1,076 patients received Restoril and 783 received placebo. The following were reported in 1% or more of patients (FDA label):

The same studies also reported depression, dry mouth, diarrhoea, abdominal discomfort, weakness, blurred vision and nightmares at rates at or near placebo. Anorexia, ataxia, loss of balance, tremor, increased dreaming, breathlessness, palpitations, vomiting, backache, sweating and burning eyes were reported at 0.5–0.9%. Amnesia, hallucinations, nystagmus and paradoxical reactions (restlessness, overstimulation, agitation) were rare (under 0.5%) (FDA label). The UK SmPC adds numbed emotions, double vision, slurred speech, muscle weakness, incontinence, urinary retention, changes in libido, skin reactions and, occasionally, blood disorders and raised liver enzymes (UK SmPC).

Serious risks

The dementia question: what the two landmark studies found

Two publicly funded BMJ studies reached different conclusions, and that disagreement is the honest state of the evidence.

• Billioti de Gage et al., 2014 (case-control; Quebec health insurance data): 1,796 people with Alzheimer's disease were compared with 7,184 controls. Benzodiazepine use starting at least five years before diagnosis was linked to Alzheimer's (adjusted OR 1.51, 95% CI 1.36–1.69). The estimate was 1.43 after adjusting for anxiety, depression and insomnia. No association was found below 91 prescribed daily doses. The link strengthened to 1.84 above 180 doses and was stronger for long-acting drugs (1.70) than short-acting ones (1.43). The authors acknowledged that benzodiazepine use "might also be an early marker of a condition associated with an increased risk of dementia". Funding came from INSERM, the University of Bordeaux, IRESP, the French Ministry of Health and Quebec's FRSQ (BMJ 2014, full text).
• Gray et al., 2016 (prospective cohort; Seattle): 3,434 people aged 65 and over, free of dementia at the start, were followed for a mean of 7.3 years. To reduce reverse causation, the most recent year of use was excluded. Dementia risk was slightly higher with minimal use (HR 1.25 for 1–30 standardised doses) but not with the highest use (HR 1.07, 0.82–1.39). Higher use was not linked to faster cognitive decline. The authors concluded the results "do not support a causal association". The study was funded by the US National Institute on Aging and the Branta Foundation. Some authors disclosed a Merck/AGS award and institutional grants from Pfizer, Amgen and Bayer (BMJ 2016, full text).

Pure City Research reading: the causal link is unproven, with an evidence grade of Insufficient. Insomnia, anxiety and low mood can be early signs of dementia, which makes reverse causation plausible. Beers 2023 lists benzodiazepines among drugs to avoid in people who already have dementia or cognitive impairment "because of adverse CNS effects" (Beers 2023). Both studies point toward avoiding long, open-ended use.

All interactions

Who should avoid temazepam

UK contraindications (do not use)

The UK SmPC lists these contraindications (UK SmPC):

• hypersensitivity to temazepam, other benzodiazepines or the excipients;
• neuromuscular respiratory weakness, including myasthenia gravis;
• acute pulmonary insufficiency, severe respiratory depression, sleep apnoea syndrome or CNS depression;
• severe hepatic insufficiency, because it may precipitate encephalopathy;
• phobic or obsessional states and chronic psychosis;
• mild anxiety states;
• acute narrow-angle glaucoma;
• use as the only treatment for depression, or for anxiety with depression, because suicide may be precipitated;
• breastfeeding.

The UK oral solution also contains sorbitol, so it must not be given to people with hereditary fructose intolerance.

Use with extra caution

• Older adults (65+): Beers 2023 says "Avoid" (strong recommendation, moderate-quality evidence) (Beers 2023). If a prescriber does use it, the US label recommends starting at 7.5 mg (FDA label). The UK SmPC says half the normal dose (5–15 mg) may be sufficient (UK SmPC).
• History of alcohol or drug misuse, or marked personality disorder: the dependence risk is higher, so avoid routine repeat prescriptions (UK SmPC). NICE lists "taking an opioid together with a benzodiazepine" and a history of drug or alcohol misuse among factors that raise the risk of dependence problems (NICE NG215).
• Chronic lung disease, and kidney or liver impairment: give with caution. Sedatives may precipitate encephalopathy in cirrhosis (UK SmPC).
• Bereavement: "psychological adjustment may be inhibited by benzodiazepines" (UK SmPC).
• Children: not recommended for insomnia (UK), and safety and effectiveness have not been established (US).

Pregnancy and breastfeeding

The UK SmPC advises avoiding regular use in pregnancy because of the risk of neonatal withdrawal. It says high doses late in pregnancy or during labour may cause the newborn to have low body temperature, floppiness and breathing problems (UK SmPC). The current US label says observational studies "do not report a clear association with benzodiazepines and major birth defects". It warns of neonatal sedation and withdrawal after use late in pregnancy, and there is a US pregnancy registry for psychiatric medicines. Temazepam passes into breast milk. The UK contraindicates it in breastfeeding, and the US label advises monitoring exposed infants for sedation, poor feeding and poor weight gain (FDA label).

Dosage and how to take it

Your prescriber sets the dose. The ranges below are the official licensed adult ranges, given for reference only. They are not advice on what anyone should take.

Sources: UK SmPC, FDA Restoril label.

How quickly it works

Temazepam works on the first night. After a 30 mg capsule, blood levels are measurable within 10–20 minutes and peak at around 1.5 hours (FDA label). The UK SmPC advises making sure you can have 7–8 hours of uninterrupted sleep after a dose, to reduce the risk of amnesia and next-day impairment (UK SmPC). The US label says that if insomnia has not improved after 7–10 days, this "may indicate the presence of a primary psychiatric and/or medical illness that should be evaluated" (FDA label).

How to stop: tapering under supervision

Both regulators say treatment should be withdrawn gradually. The UK SmPC says treatment "should always be tapered off gradually", and that people who have taken benzodiazepines for a long time "may require a longer period during which doses are reduced" (UK SmPC). The FDA says "no standard benzodiazepine tapering schedule is suitable for all patients". It asks prescribers to create a patient-specific plan and to pause or slow the taper if withdrawal reactions appear (FDA 2020, FDA label).

NICE's 2022 guideline on dependence-forming medicines sets out the UK approach (NICE NG215):

• Do not stop abruptly unless there are exceptional medical circumstances (recommendation 1.5.6).
• Use "a slow, stepwise rate of reduction proportionate to the existing dose, so that decrements become smaller as the dose is lowered" (1.5.11).
• "If the person is withdrawing from a benzodiazepine with a short half-life, consider switching to a benzodiazepine with a longer half-life" (1.5.13).
• Agree a pace the person finds acceptable, allow intolerable symptoms to settle before the next step, and consider giving the person some control over the size of the decrements (1.5.10).
• Explain that "withdrawal can be difficult, and may take several months or more" (1.5.9).

The long-standing UK clinical reference is Professor Heather Ashton's manual, based on her benzodiazepine withdrawal clinic, which opened in 1982. It recommends switching people on shorter-acting drugs to diazepam, one dose at a time, and then tapering slowly. As "a very rough guide" for a person on 40 mg diazepam a day or its equivalent, it describes reductions of 2 mg every 1–2 weeks down to 20 mg, then 1 mg every week or two. That gives a total withdrawal of 30–60 weeks, with the pace adjusted to the individual (Ashton Manual, ch. II). The manual reflects clinical experience rather than trials, and people who used temazepam only briefly will usually need a much shorter taper. That decision belongs to the prescriber.

What helps people stop long-term use? A Cochrane review found that CBT plus a taper was more likely than a taper alone to lead to stopping within four weeks after treatment (RR 1.40) and at three months (RR 1.51). The benefit was not sustained at six months and beyond (Darker et al., Cochrane 2015). In the EMPOWER trial, older long-term users were mailed an education booklet on the risks with a stepwise tapering protocol. At six months, 27% of them had stopped, compared with 5% of controls (number needed to treat, 4) (Tannenbaum et al., JAMA Intern Med 2014). Evidence for drugs that help withdrawal is low or very low quality. The Cochrane authors could not draw firm conclusions, and they called for future trials to be run "independently of industry involvement" (Baandrup et al., Cochrane 2018).

Follow the money: who makes it and who funded the evidence

Who makes and sells it

• US brand (Restoril): the NDA holder is SpecGx LLC of Webster Groves, Missouri (Drugs@FDA). FDA correspondence shows the NDA was held by Mallinckrodt Inc. in 2002 and 2004, and the 2008 label called Restoril a trademark of Mallinckrodt Inc. (FDA 2002, 2008 label). The 2026 label carries "© 2026 Par Health, Inc." (FDA label). According to a 10 November 2025 announcement, Par Health is an "independent, private company" headquartered in St. Louis, Missouri. It was spun off from Mallinckrodt plc and combines the generics portfolios of Mallinckrodt and Endo following their July 2025 merger (Par Health / SpecGx newsroom).
• Original developer: not verified. The Drugs@FDA page does not name the 1981 applicant. A reference to "SandoPak" packaging in the 2002 FDA letter hints at a Sandoz lineage, but we could not confirm it from a primary source.
• UK: the one temazepam product with an SmPC on emc is an oral solution licensed to Syri Limited (trading as Thame Laboratories / SyriMed), Ruislip, UK. It was first authorised on 20 October 2017 (UK SmPC). Other UK tablet manufacturers were not verified from fetched pages.
• Revenue: we found no primary source giving temazepam or Restoril sales. As an old generic, it is a low-price product. NICE's 2004 advice to prescribe whichever of zolpidem, zopiclone or short-acting benzodiazepines has "the lowest purchase cost" reflects this (NICE TA77).

Who funded the evidence

• Pivotal approval trials: the Restoril label describes 2-week placebo-controlled sleep-lab studies in chronic insomnia, plus 1,076 treated patients in controlled studies. The label does not state who ran or funded them. Since they supported a company's 1981 new-drug application, they should be treated as sponsor-generated evidence, and they are not publicly available in full (FDA label). This is an inference from the regulatory route, not a documented funding statement.
• Independent reviews: the most informative efficacy-and-harm syntheses come from public or unfunded sources:
  • Glass 2005: "Funding: None".
  • De Crescenzo 2022: UK NIHR. Some authors have ties to Boehringer Ingelheim, Angelini and Janssen, which are unrelated to temazepam.
  • Holbrook 2000: part-funded by the Canadian Pharmaceutical Association and the Canadian Medical Association.
  • Morin 1999: NIMH.
  • Billioti de Gage 2014: French and Quebec public bodies.
  • Gray 2016: US NIA.
• Guidelines: AASM funded its own 2017 guideline. Its authors declared consultancies with, among others, Merck, Janssen, Jazz, Lundbeck and Purdue Pharma. One author, who served on Merck's advisory board, did not take part in the suvorexant recommendation (AASM 2017). As an organisation, AASM discloses industry support. For 2023 this included $60,000 each from Idorsia, Jazz, Avadel and Zoll and $20,000 from Eisai through its Industry Engagement Program. These are makers of newer sleep drugs and devices, not temazepam (AASM disclosure). Beers 2023 had "no sponsor" (Beers 2023).

Documented integrity events

We found no regulator enforcement action or court finding specific to temazepam or Restoril in the pages fetched for this review. The FDA's 2020 action was a class-wide labelling requirement and did not concern any one company's conduct (FDA 2020). Litigation involving the parent companies over other products (opioids) exists, but it was not verified from primary sources in this review and is not about temazepam, so it is not detailed here.

UK prescribing context

Public Health England's 2019 review of dependence-forming prescribed medicines found that 1.4 million adults in England (3%) received a benzodiazepine prescription in 2017 to 2018. About 120,000 received one continuously from April 2015 to March 2018. It found that "use of shorter-acting benzodiazepines" was associated with long-term use, and it noted that guidelines specify benzodiazepines "should not usually be prescribed for longer than 2 to 4 weeks" (PHE Prescribed medicines review, summary).

Related research

• Sleep prevention guide: evidence-based ways to sleep better
• Sleep supplements: what the independent evidence shows
• Melatonin: independent evidence review
• Magnesium: independent evidence review
• L-theanine: independent evidence review
• Ashwagandha: evidence and safety
• Stress, anxiety and depression prevention guide
• Zopiclone: independent evidence
• Zolpidem: independent evidence
• Daridorexant: independent evidence
• Trazodone: independent evidence

Frequently asked questions

Sources and funding notes

• Temazepam 10mg/5ml Oral Solution, Summary of Product Characteristics (emc): licence holder Syri Ltd (UK); regulator-approved manufacturer document; text revised 31/08/2022.
• Restoril (temazepam) US prescribing information, via openFDA: SpecGx LLC / Par Health (US); FDA-approved label, effective June 2026.
• Drugs@FDA, NDA 018163: US regulator; approval 27 February 1981; supplement history.
• FDA letter to Mallinckrodt, S-053 (2002), Restoril label 2008 and FDA approval letter S-067 (2021): US regulator correspondence showing ownership history and the boxed-warning update.
• FDA Drug Safety Communication, 23 September 2020: US regulator; no sponsor.
• MHRA Drug Safety Update, 18 March 2020: UK regulator; no sponsor.
• NICE TA77 (2004): UK public body; guidance on hypnotics for short-term insomnia.
• NICE NG215 (2022), medicines associated with dependence or withdrawal symptoms: UK public body.
• NHS: Insomnia (reviewed 19 March 2024): UK public health service. The NHS temazepam page is no longer available.
• Misuse of Drugs Regulations 2001, Schedule 3 and Drug Driving (Specified Limits) Regulations 2014: UK legislation.
• GOV.UK: Drugs and driving: the law: UK government.
• Sateia et al., AASM clinical practice guideline, J Clin Sleep Med 2017: funded by AASM (US); author consultancies declared (Merck, Janssen, Jazz, Lundbeck, Purdue and others).
• AASM CMSS financial disclosure (2023): industry support to the society; none from temazepam makers is listed.
• Riemann et al., European Insomnia Guideline 2023, J Sleep Res: European Sleep Research Society; Europe PMC lists NIHR grant 16/84/01; author COIs not verified.
• De Crescenzo et al., Lancet 2022: UK NIHR Oxford Health BRC; first author employed by Boehringer Ingelheim; some authors had Angelini fees; senior author leads Janssen-sponsored seltorexant trials.
• Glass et al., BMJ 2005: University of Toronto (Canada); "Funding: None"; no competing interests.
• Holbrook et al., CMAJ 2000: McMaster University (Canada); part-funded by the Canadian Pharmaceutical Association and Canadian Medical Association (CMAJ full text); no competing interests declared.
• Morin et al., JAMA 1999: US NIMH grant MH47020; academic (Canada/US).
• American Geriatrics Society Beers Criteria 2023: "no sponsor"; some panel consultancies declared (US).
• Billioti de Gage et al., BMJ 2014: INSERM, University of Bordeaux, IRESP, French Ministry of Health, Quebec FRSQ; some authors had unrelated industry honoraria (France/Canada).
• Gray et al., BMJ 2016: US National Institute on Aging and Branta Foundation; some authors disclosed a Merck/AGS award and institutional Pfizer, Amgen and Bayer grants (US).
• Darker et al., Cochrane 2015: part-funded by Health Research Board Ireland; no author conflicts.
• Baandrup et al., Cochrane 2018: Denmark; some authors declare industry lecture and advisory fees; the lead author ran one included trial, with independent data extraction.
• Tannenbaum et al., EMPOWER trial, JAMA Intern Med 2014: Canadian Institutes of Health Research (Canada).
• Public Health England, Prescribed medicines review (2019): UK government agency.
• Ashton CH, Benzodiazepines: How They Work and How to Withdraw (2002), ch. II: Newcastle University clinical pharmacologist (UK); no commercial sponsor; clinical-experience guidance.
• Par Health spin-off announcement, 10 November 2025: company press release (US); used only for ownership facts.

Last reviewed: 26 September 2026. Educational information from Pure City Research, not medical advice. Medicine decisions belong with a qualified prescriber.