Suvorexant (Belsomra): Independent Evidence on Insomnia, Side Effects & Risks

by
Key takeaways
  • Suvorexant (Belsomra) was the first orexin receptor antagonist sleeping pill. Merck discovered it and the US FDA approved it on 13 August 2014. It blocks the brain's "stay awake" orexin signal instead of boosting sedating GABA signalling like benzodiazepines and Z-drugs do (FDA approval package, Cox et al., J Med Chem 2010).
  • It is not licensed in the UK. A search of the UK electronic Medicines Compendium returns no suvorexant product, and the only orexin antagonist approved by the European Medicines Agency is daridorexant (EMC search, Actas Esp Psiquiatr review 2024).
  • The benefits are real but modest. In Merck's two 3-month trials, suvorexant 15–20 mg beat placebo on night-time wakefulness by 17–31 minutes. On time to fall asleep, the gap was 0–10 minutes, and in one trial it had disappeared by month 3 (FDA label).
  • The approved dose came from a fight between the FDA and Merck. Merck asked for 15–40 mg. The FDA refused to approve it until a 10 mg starting tablet existed, citing next-day driving risk and dose-related suicidal ideation. Both the FDA and its advisers judged the 30/40 mg doses too risky for too little extra benefit (FDA medical review, 2013–14).
  • The US sleep specialists' guideline (AASM 2017) only "suggests" suvorexant, and only for staying asleep. That is a WEAK recommendation based on low-quality evidence. One author recused himself from the suvorexant recommendation because he sat on a Merck advisory board (Sateia et al., JCSM 2017).
  • The largest publicly funded comparison, the Lancet 2022 network meta-analysis, found suvorexant can work for short-term treatment. However, it grouped suvorexant with drugs that have poor tolerability or no long-term data (De Crescenzo et al., Lancet 2022).
  • Evidence grade: Moderate for short-term sleep maintenance. Weak for sleep onset and long-term use. Almost all efficacy trials were paid for by Merck.

Independent evidence review · Prescription medicine

Suvorexant is a genuinely new kind of sleeping pill, but it is not a stronger one. Merck's own trials show it mainly helps people stay asleep, adding roughly 11–22 minutes of patient-reported sleep per night more than placebo. Its main risks are next-day drowsiness and impaired driving. Its approved dose exists largely because FDA reviewers insisted on a lower one. Every major guideline still puts cognitive behavioural therapy for insomnia (CBT-I) first (European Insomnia Guideline 2023, ACP 2016). In the UK, suvorexant is not an option at all; NHS doctors "now rarely prescribe sleeping pills" (NHS insomnia page).

Best evidence for short-term help with staying asleep (sleep maintenance insomnia) in adults
Main risks next-day sleepiness and driving impairment, sleep paralysis, dose-related suicidal ideation, interactions with CYP3A inhibitors
Key rule lowest effective dose (10 mg start in the US), at least 7 hours in bed, no alcohol
Safety first
Suvorexant is a prescription-only, controlled (Schedule IV in the US) sleep medicine. Do not start it, stop it or change the dose without your prescriber. It can impair driving the next day even if you feel awake. The label warns patients on 20 mg against next-day driving, and warns that alcohol, opioids, benzodiazepines and other sedatives add to its effects. Do not combine it with alcohol. If you notice worsening depression or thoughts of suicide, or if you sleep-walk or sleep-drive, stop and seek urgent medical help. In an emergency, call your local emergency number (FDA label, section 5).

Table of contents

Evidence summary

Claim Evidence Source Funding / conflict Strength
Helps people stay asleep (sleep maintenance) Two 3-month RCTs. On polysomnography, 15–20 mg cut wake after sleep onset by 17–31 minutes more than placebo. Patient-estimated total sleep time rose by 11–22 minutes more than placebo. FDA label, Tables 5–6; Herring et al., Biol Psychiatry 2016 Merck-sponsored registration trials; lead author a Merck employee. Figures re-checked by FDA reviewers. Moderate
Helps people fall asleep (sleep onset) Objective sleep-onset gains over placebo ranged from 0 to 10 minutes. In one trial the difference was 0 minutes at month 3. AASM judged the 15/20 mg latency gains as failing to meet clinical significance. FDA label, Table 3; AASM 2017 Merck trials; AASM guideline funded by AASM Weak
Works across the drug class in head-to-head rankings A network meta-analysis of 154 double-blind RCTs listed suvorexant among licensed drugs that "can be effective in the acute treatment of insomnia but are associated with poor tolerability, or information about long-term effects is not available". De Crescenzo et al., Lancet 2022 Funded by UK NIHR (public). First author employed by Boehringer Ingelheim; senior author runs Janssen trials of seltorexant, a rival orexin drug. Moderate (acute)
Long-term use (1 year) At 40/30 mg (above today's maximum dose), 62% of patients on suvorexant and 63% on placebo completed 1 year. Somnolence affected 13% vs 3%. Efficacy was formally tested only over the first month. Michelson et al., Lancet Neurol 2014 Funded by Merck; authors Merck employees Weak
Insomnia in Alzheimer's disease In a 4-week RCT (n=285), total sleep time improved by 28 minutes more than placebo (95% CI 11–45). This result is now in the US label. Herring et al., Alzheimers Dement 2020 Merck-funded; most authors Merck employees Moderate (short-term)
Prevents delirium in hospital A Japanese RCT found delirium in 16.8% vs 26.5% of patients, a difference that was not statistically significant (P=.13). A pooled meta-analysis of the drug class shows a possible benefit, rated low certainty. Hatta et al., JAMA Netw Open 2024; de Oliveira et al., Crit Care Med 2025 The RCT was funded by MSD K.K. (Merck's Japanese arm). The meta-analysis reported no conflicts. Insufficient
Next-day driving impairment On average, next-morning driving was not impaired. However, some people on 20 mg were impaired, and 4 women on suvorexant stopped driving tests early because of sleepiness. The label warns 20 mg users against next-day driving. Vermeeren et al., Sleep 2015; FDA label 14.2 University of Maastricht study within Merck's programme; the companion elderly study discloses Merck funding. Risk
Narcolepsy-like side effects across the class Across 11 RCTs (7,703 patients) of orexin antagonists, excessive daytime sleepiness RR 2.15 and sleep paralysis RR 3.40 vs placebo. Na et al., Sleep 2024 Funded by the National Research Foundation of Korea (public) Risk

Independent evidence and credibility scorecard

Source Who funds it Country Independence tier (1–4) Credibility (A–D) Why it would tell the truth / residual bias
FDA medical and decisional reviews US government. About 77% of the drug-review budget comes from industry user fees (FDA PDUFA report FY2025). USA 2 A− Reviewers have the raw trial data and a legal duty to be accurate, and in this case they pushed back on the company's dose. The residual bias is the user-fee dependence and the fact that reviewers can only judge what the company submits.
De Crescenzo et al., Lancet 2022 UK NIHR UK / Italy 1–2 A− A pre-registered analysis that included unpublished trial data. The residual bias is that the author declarations include a Boehringer Ingelheim employee and an investigator on Janssen seltorexant trials.
AASM 2017 guideline AASM (a professional society). The AASM itself discloses company support, including from orexin-drug makers (AASM disclosure). USA 2–3 B It uses the GRADE method, and the Merck-tied author recused himself. The residual bias is the society-level industry income. Its evidence base was only two Merck trials.
European Insomnia Guideline 2023 European Sleep Research Society; an NIHR grant is listed Pan-European 2 B A consensus of 44 authors. Author conflicts could not be verified because the publisher blocked access.
Na et al., Sleep 2024; Pan et al., 2024; Araujo et al., 2026 Korean public grant, or no external funding / no conflicts declared Korea / China / Brazil 1 B These are academic meta-analyses with no sponsor. The residual bias is that they pool trials that were almost all run by the manufacturers.
Herring et al. 2016, Michelson et al. 2014, Herring et al. 2012 Merck USA (multinational sites) 4 B for data, C for framing These trials were registered and audited by the FDA; four sites were inspected with no action needed. The residual bias is that the company designed, ran and wrote up the trials and chose the doses tested.
Merck 10-K FY2025 Merck USA 4 B for figures Securities law penalises misstated revenue. The filing contains no clinical claims.

What suvorexant is

Suvorexant is a dual orexin receptor antagonist (DORA), sold as Belsomra. The FDA described it at approval as "both a new molecular entity and first-in-class" (FDA Office Director decisional memo). The molecule, coded MK-4305, was designed by Merck Research Laboratories in West Point, Pennsylvania (Cox et al., J Med Chem 2010). Merck filed the investigational new drug application on 10 April 2008 (FDA memo).

The FDA approved Belsomra (NDA 204569) on 13 August 2014, after first rejecting the application in June 2013 (FDA approval letter). Merck's own pooled analysis says suvorexant is approved in the US, Japan and Australia (Herring et al., Sleep Med 2019). The US label is held by Merck Sharp & Dohme LLC, Rahway, New Jersey (DailyMed).

US status: prescription only, and a Schedule IV controlled substance (label section 9). No generic is on sale. FDA records show one generic application, from Hetero Labs, with only tentative approval dated 22 December 2025 and a marketing status of "None" (openFDA Drugs@FDA).

UK and EU status: not licensed. The UK electronic Medicines Compendium returns no results for suvorexant (EMC). A 2024 review notes that daridorexant is the only orexin antagonist approved by the European Medicines Agency (Actas Esp Psiquiatr 2024). In the UK, the orexin antagonist NICE has appraised is daridorexant, not suvorexant (NICE TA922). We found no public statement explaining why Merck has not sought a UK or EU licence, and we do not speculate.

How it works

Orexins (also called hypocretins) are brain peptides that promote wakefulness. The FDA's decisional memo summarises the history: in the late 1990s, a mutation of an orexin receptor was found to cause narcolepsy in dogs, and mice lacking the orexin gene were found to have narcolepsy (FDA memo). Suvorexant binds both orexin receptors, OX1R and OX2R (Ki 0.55 and 0.35 nM). Blocking orexin A and B "is thought to suppress wake drive" (label 12.1–12.2).

This differs from benzodiazepines and Z-drugs such as zolpidem, which amplify the brain's main inhibitory (GABA) system. The label notes the downside of the orexin approach: blocking orexin "may also underlie potential adverse effects such as signs of narcolepsy/cataplexy". In dogs, Merck's toxicology studies found cataplexy-like behaviour near peak blood levels (FDA memo).

Pharmacokinetics.

  • Peak levels come about 2 hours after a dose (range 30 minutes to 6 hours). A high-fat meal delays the peak by about 1.5 hours.
  • The half-life is about 12 hours, so a meaningful amount is still in the blood the next morning.
  • It is cleared mainly by the liver enzyme CYP3A.
  • Exposure is higher in women and in obese people. Obese women have 46% higher total exposure (AUC) than non-obese women.

Source for all of the above: label 12.3.

What it is prescribed for

Licensed use (US): "treatment of insomnia characterized by difficulties with sleep onset and/or sleep maintenance" in adults (label section 1). There is no UK licence.

Where guidelines place it:

  • Every major guideline puts CBT-I first. The American College of Physicians (ACP) gives CBT-I a strong recommendation; adding drugs only after CBT-I fails is a weak recommendation based on low-quality evidence (ACP 2016). The European guideline recommends CBT-I first-line "in adults of any age" (grade A) (Riemann et al., 2023). NICE calls CBT-I "the standard first treatment" (NICE TA922).
  • AASM 2017 (US): "We suggest that clinicians use suvorexant as a treatment for sleep maintenance insomnia (versus no treatment) in adults. (WEAK)". This recommendation does not cover sleep onset. The overall quality of evidence was "low due to imprecision and risk of publication bias" (Sateia et al., 2017).
  • European guideline 2023: "Orexin receptor antagonists can be used for periods of up to 3 months or longer in some cases" (grade A) (Riemann et al., 2023). In practice, suvorexant is not available in Europe.
  • AASM 2026 combination guideline: suggests combining CBT-I with insomnia medication rather than using medication alone. It also suggests against adding medication to CBT-I compared with CBT-I alone. Both are conditional recommendations based on low-certainty evidence (Buysse et al., 2026).

What works and what does not

Use Verdict What the evidence shows
Staying asleep (fewer minutes awake at night), short-term Works Consistent reduction in wake time on polysomnography. The AASM rated the reduction in wake after sleep onset as clinically significant (AASM).
Falling asleep faster Mixed Small and inconsistent effect at approved doses; 0 minutes vs placebo at month 3 in Study 2 (label Table 3).
Total sleep time and insomnia severity Mixed Statistically better than placebo. Patient-reported gains were mostly below the AASM clinical-significance thresholds, and the placebo response was large (Herring 2019).
Insomnia in mild to moderate Alzheimer's disease Works (4 weeks) +28 minutes of total sleep time vs placebo in one Merck RCT (Herring 2020).
Long-term (beyond 3 months) benefit Insufficient The only 1-year RCT used higher-than-approved doses and formally tested efficacy only in month 1 (Michelson 2014).
Delirium prevention in hospital Insufficient The largest RCT was not statistically significant, and the class-level meta-analysis is low certainty (Hatta 2024, de Oliveira 2025).
Better than CBT-I Insufficient We found no head-to-head trial. Guidelines place CBT-I first (European guideline).

Benefits by claim

The pivotal trials, in numbers

Merck ran two similar 3-month RCTs, called Study 1 and Study 2 in the label. They enrolled 1,021 and 1,019 patients (Herring et al., 2016). The trials were designed around a higher dose (40 mg for non-elderly adults, 30 mg for elderly adults). Fewer patients were randomised to the 20/15 mg doses that ended up close to the approved range. The FDA label reports the 15–20 mg results against placebo (label Tables 3–6):

Outcome (difference vs placebo) Study 1, month 1 Study 1, month 3 Study 2, month 1 Study 2, month 3
Time to sleep onset, polysomnography (minutes)−10−8−80
Time to sleep onset, patient-estimated (minutes)−5−5−7−8
Wake after sleep onset, polysomnography (minutes)−26−17−24−31
Total sleep time, patient-estimated (minutes)+16+11+21+22

Placebo did much of the work. In Study 1, patient-estimated total sleep time rose by 41 minutes on placebo by month 3, against 51 minutes on suvorexant (label Table 6). That fits a wider pattern. An independent meta-analysis found that 63.56% of the drug response in hypnotic trials also occurred in placebo groups (Winkler & Rief, Sleep 2015).

The AASM re-analysis applied clinical-significance thresholds (Sateia et al., 2017):

  • Sleep latency: 10 mg reduced it by 2.3 minutes (95% CI 13.68 lower to 9.08 higher), which is not clinically significant. 20 mg reduced it by 22.3 minutes, which is clinically significant. The 15/20 mg arm of Herring 2016 trial 1 reduced it by 8.1 minutes, which is not.
  • Wake after sleep onset: reduced by 21.4 minutes (10 mg) and 28.1 minutes (20 mg), both clinically significant.
  • Subjective total sleep time: +10.6 minutes, below the threshold. Sleep quality ratings "showed minimal change".

Insomnia Severity Index

In Merck's pooled analysis, Insomnia Severity Index scores fell by 6.2 points at month 3 on 20/15 mg, against 4.9 on placebo. That is a gap of 1.3 points on a 28-point scale. Response, defined as a 6-point improvement or more, was seen in 55.5% vs 42.2% (Herring et al., Sleep Med 2019). This was an exploratory outcome in Merck-funded trials.

One year of use

In the 1-year Merck trial (40 mg for under-65s, 30 mg for older adults, both above today's 20 mg maximum):

  • At month 1, subjective total sleep time rose by 38.7 vs 16.0 minutes (difference 22.7, 95% CI 16.4–29.0).
  • Time to sleep onset fell by 18.0 vs 8.4 minutes (difference −9.5).
  • Serious adverse events occurred in 5% of patients on suvorexant and 7% on placebo.

The primary objective was safety, and efficacy was tested formally only over the first month (Michelson et al., Lancet Neurol 2014). Independent pooled analyses of long-term DORA trials (6 RCTs, 3,546 people) found sustained improvements in patient-reported sleep. Adverse events were similar to placebo at 6 months but higher at 12 months (Araujo et al., 2026).

How it compares with other sleeping pills

The Lancet 2022 network meta-analysis did not place suvorexant among its best-profile drugs. Those were eszopiclone and lemborexant, with the caveats that "eszopiclone might cause substantial adverse events and safety data on lemborexant were inconclusive". It did find that zopiclone caused more dropouts than suvorexant because of side effects (OR 3.13, 95% CI 1.47–6.67; low certainty) (De Crescenzo et al., 2022). No trial compared suvorexant directly with another sleeping pill. See our lemborexant and daridorexant reviews for the sister drugs.

The dose debate, from the FDA's own files. Merck asked the FDA to approve starting doses of 20 mg (under 65) and 15 mg (65 and over), with 40/30 mg for non-responders. At the 22 May 2013 advisory committee, members voted as follows:

  • 13–3 (1 abstention) that the 15/20 mg starting doses were acceptably safe.
  • 7 yes, 8 no (2 abstentions) on whether the 30/40 mg doses were acceptably safe.
  • 12–4 that the doses worked for sleep onset, and 16–0 that they worked for sleep maintenance.

Committee members also noted that Merck "never established 10 mg as an effective dose". Instead, "it was FDA that performed the post hoc analysis of an underpowered Phase 2 trial" suggesting it worked.

FDA neurology director Russell Katz wrote that, "given the risks of next day effects at even the 15 and 20 mg dose groups, especially next-day driving impairment", suvorexant "cannot be marketed with an acceptable safety profile without the availability of the 10 mg dose". The FDA issued a Complete Response letter on 28 June 2013. It asked for a 10 mg tablet and a 5 mg tablet for high-exposure groups such as obese women, and concluded that 30 and 40 mg "should not be marketed". Merck resubmitted on 14 February 2014, and the approved label starts at 10 mg with a 20 mg maximum (FDA medical review, FDA approval letter).

What this means: before approval, the fixed 10 mg starting dose had been tested for insomnia in only 62 patients, in a 4-week phase 2 crossover trial (Herring et al., Neurology 2012). The later Alzheimer's trial also started patients at 10 mg, and 77% were increased to 20 mg (label 14.1).

Risks and all side effects

Belsomra has no boxed warning. The 2019 boxed warning for complex sleep behaviours applies to eszopiclone, zaleplon and zolpidem, not to suvorexant. The suvorexant label does, however, list complex sleep behaviours among its warnings (label section 5).

Effect How often / detail Severity
Next-day impairment and impaired driving Can occur "even when used as prescribed" and "may not be reliably detected by ordinary clinical exam". Effects can persist "for up to several days" after stopping. Patients on 20 mg should be cautioned against next-day driving (label 5.1). High
Worsening depression / suicidal ideation "A dose-dependent increase in suicidal ideation was observed" (label 5.2). In 12-month data, 8 of 1,268 patients on suvorexant (all at the higher doses) vs 0 of 1,012 on placebo had suicidal ideation classified by protocol. The trials largely excluded people with depression (FDA review). High
Complex sleep behaviours Sleep-walking, sleep-driving, and eating, phoning or having sex while not fully awake. "Discontinue BELSOMRA immediately" if this happens (label 5.3). High
Sleep paralysis, hypnagogic/hypnopompic hallucinations, cataplexy-like leg weakness Risk increases with dose (label 5.4). Across the orexin-antagonist class, sleep paralysis RR 3.40 vs placebo (Na et al., 2024). Moderate
Somnolence 7% vs 3% on placebo at 15/20 mg; 8% in women vs 3% in men. Dose-related: 2% at 10 mg, 5% at 20 mg, 12% at 40 mg (label 6.1). Moderate
Falls (older adults) The label warns of higher fall risk, particularly in elderly people. In the Alzheimer's trial, falls occurred in 2% vs 0% on placebo (label 5.1, 6.1). A meta-analysis found no significant increase in falls or fractures, but data are limited (Pan et al., 2024). Moderate
Breathing in severe sleep apnoea or COPD Not studied in severe obstructive sleep apnoea or severe COPD. Effects in mild-to-moderate disease "cannot be excluded" (label 5.5, 8.6). Moderate
Headache, dizziness, abnormal dreams, dry mouth, diarrhoea, cough, upper respiratory infection Headache 7% vs 6%; dizziness 3% vs 2%; the others 2% vs 1% (label Table 2). Mild
Cholesterol Small dose-related rise: +2 mg/dL at 20 mg vs −4 mg/dL on placebo after 4 weeks (label 6.1). Mild
Post-marketing reports Palpitations, tachycardia, nausea, vomiting, psychomotor hyperactivity, anxiety and itching. Causality is uncertain (label 6.2). Variable

Dependence, abuse and withdrawal.

  • Suvorexant is Schedule IV in the US.
  • In recreational drug users, 40–150 mg produced "drug liking" similar to zolpidem 15–30 mg.
  • In trials, there was "no evidence for physical dependence" and no reported withdrawal symptoms.
  • No clear rebound insomnia was seen after stopping (label 9, 14.2).

Source: FDA label. These are manufacturer trial findings over months, not decades.

Regulators outside the US. Suvorexant is not authorised in the UK or EU, so there are no MHRA or EMA safety communications specific to it.

All interactions

Interacting drug or substance Examples (from the label) Effect Label action
Strong CYP3A inhibitors Ketoconazole, itraconazole, posaconazole, clarithromycin, nefazodone, ritonavir, saquinavir, nelfinavir, indinavir, boceprevir, telaprevir, telithromycin, conivaptan Large rise in suvorexant levels Not recommended
Alcohol Any alcoholic drink Additive psychomotor impairment Do not combine
Other CNS depressants Benzodiazepines, opioids, tricyclic antidepressants, other sleeping pills Additive sedation, next-day impairment Dose reduction may be needed; use with other insomnia drugs not recommended
Moderate CYP3A inhibitors Amprenavir, aprepitant, atazanavir, ciprofloxacin, diltiazem, erythromycin, fluconazole, fosamprenavir, grapefruit juice, imatinib, verapamil Raised suvorexant levels 5 mg dose; generally no more than 10 mg
Strong CYP3A inducers Rifampin, carbamazepine, phenytoin Suvorexant levels fall substantially; may not work Efficacy may be reduced
Digoxin Digoxin Slightly higher digoxin levels (intestinal P-gp inhibition) Monitor digoxin
Paroxetine Paroxetine 20 mg No clinically significant interaction in a healthy-volunteer study No change

Source: FDA label sections 2.4, 7 and 12.3. The suvorexant label does not name St John's wort. It is a known CYP3A inducer, and the sister drug lemborexant's label lists it as a strong inducer. Ask a pharmacist before combining any herbal product; see our St John's wort review.

Who should avoid suvorexant

  • People with narcolepsy: this is the label's only formal contraindication (label section 4).
  • Severe liver impairment: not studied and not recommended. In moderate impairment, the half-life rose from about 15 to about 19 hours. No adjustment is needed for kidney impairment (label 8.7–8.8).
  • Severe sleep apnoea or severe COPD: not studied (label 5.5).
  • People with depression or suicidal thoughts: the label advises immediate evaluation of suicidal ideation and prescribing the smallest feasible number of tablets. The FDA noted that the trials largely excluded this group (FDA review).
  • People with a history of substance misuse: the label advises careful follow-up (label 9.2).
  • Pregnancy: postmarketing human data are "insufficient to establish a drug-associated risk". Animal studies showed effects only at high multiples of the human dose (label 8.1).
  • Breastfeeding: suvorexant passes into breast milk at a relative infant dose below 1%. There are no data on effects in breastfed infants (label 8.2).
  • Older adults: the label reports no clinically meaningful differences in safety or effectiveness at recommended doses, but warns of higher fall risk (label 8.5). We could not confirm from the sources we accessed how the 2023 AGS Beers Criteria rate orexin antagonists. Beers says to avoid benzodiazepines and Z-drugs.
  • Children: safety and effectiveness have not been established (label 8.4).

Dosage and how to take it

Your prescriber sets the dose. The official US adult dosing is summarised below for reference, not as personal advice (label section 2):

  • Recommended dose: 10 mg, no more than once a night, taken within 30 minutes of going to bed, with at least 7 hours before the planned wake time.
  • Maximum: 20 mg once nightly, only if 10 mg is well tolerated but not effective. "Use the lowest dose effective for the patient."
  • With moderate CYP3A inhibitors: 5 mg, generally no more than 10 mg. It is not recommended with strong inhibitors.
  • Food: taking it with or soon after a meal may delay its effect.
  • Obese women: consider the higher exposure before increasing the dose.

How long before it works: the label reports effects on night 1 (objective) and week 1 (subjective) that were "generally consistent with later time points". If insomnia has not improved after 7–10 days, the label advises re-evaluation for another medical or psychiatric cause (label 5.6, 14.1).

How to stop: Merck's trials found no clear withdrawal or rebound insomnia after stopping (label 14.2). Stopping should still be planned with your prescriber, especially if you have taken it for a long time or alongside other sedatives. If you are stopping because of a complex sleep behaviour, the label says to stop immediately and seek medical advice.

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

Originator and owner. Merck & Co (known as MSD outside the US and Canada) discovered suvorexant and still owns and markets it. Merck is headquartered in Rahway, New Jersey (Cox et al., 2010, DailyMed). In Japan, the company's MSD K.K. subsidiary funded the recent delirium trial (Hatta et al., 2024).

Revenue. Merck's annual report (form 10-K) lists Belsomra worldwide sales as follows (Merck 10-K FY2025):

YearUSRest of worldTotal
2023$81m$150m$231m
2024$72m$150m$222m
2025$82m$104m$186m

Most sales are outside the US. The 10-K does not break down sales by country. For comparison, Eisai reported its rival lemborexant (Dayvigo) at ¥64.3 billion in the year to March 2026 (see our lemborexant review).

Generics. None are marketed in the US. One generic application (Hetero Labs, an Indian company) holds tentative approval only (openFDA).

Who paid for the evidence.

  • Merck-funded: the phase 2 dose-finding trial (Herring 2012), both pivotal 3-month trials (Herring 2016), the 1-year trial (Michelson 2014, "Funding: Merck & Co Inc"), the Alzheimer's trial (Herring 2020), the elderly driving study, run with Maastricht University and SGS Life Sciences (Vermeeren 2016), and the Japanese delirium trial, where MSD K.K. helped design the study, collect and analyse data and prepare the manuscript (Hatta 2024).
  • Publicly funded or unfunded: the Lancet network meta-analysis (UK NIHR) (De Crescenzo 2022), the Korean-government-funded safety meta-analysis (Na 2024), and the falls, long-term and delirium meta-analyses, which declared no conflicts (Pan 2024, Araujo 2026, de Oliveira 2025). One independent elderly meta-analysis declared no funding but obtained some raw data "via an agreement with Merck" (Rollo et al., Sleep 2026).

Guideline conflicts.

  • AASM 2017: funded by the AASM. Dr Andrew Krystal "serves on a scientific advisory board for Merck, and therefore did not participate in the development of the suvorexant recommendation". He also disclosed Merck consulting (AASM 2017 disclosure statement). The guideline itself notes that quality ratings were downgraded partly "due to the funding source for most pharmacological clinical trials and the attendant risk of publication bias".
  • AASM 2026 combination guideline: its lead author consults for Eisai and Idorsia, the makers of the rival orexin drugs (Buysse et al., 2026).
  • AASM as an organisation: it discloses company support from sleep-drug makers under its industry programme (AASM disclosure).

The regulator. In FY2025, about 77% of the cost of FDA human-drug review was paid for by industry user fees (FDA PDUFA financial report). The suvorexant file is nonetheless a documented case of FDA reviewers overruling the company's preferred doses. No regulatory integrity event (such as a fine or data-misconduct finding) involving suvorexant was found in the sources we reviewed.

Frequently asked questions

How long does suvorexant take to work?

It reaches peak blood levels about 2 hours after a dose (range 30 minutes to 6 hours), and a meal can delay this. That is why the label says to take it within 30 minutes of going to bed. In trials, effects on night 1 were generally consistent with later time points. If insomnia has not improved after 7–10 days, the label advises re-evaluation (FDA label).

Can you drink alcohol on suvorexant?

No. The label says patients "should be advised not to consume alcohol in combination with BELSOMRA because of additive effects". A study showed additive psychomotor impairment when the two were combined (FDA label 5.1, 7.1).

Can I drive the morning after taking Belsomra?

Be cautious. In Merck's driving studies, some people on 20 mg were impaired about 9 hours after dosing, and four women stopped their driving tests early because of sleepiness. The label cautions 20 mg users against next-day driving, and warns that people on lower doses can also be affected because sensitivity varies (Vermeeren et al., 2015, FDA label).

Is suvorexant addictive?

It is a Schedule IV controlled substance in the US. In recreational drug users, high doses produced "drug liking" similar to zolpidem. In the clinical trials, however, there was no evidence of physical dependence and no reported withdrawal symptoms (FDA label section 9).

How do I stop suvorexant safely?

Talk to your prescriber first. Merck's 3-month trials found no clear withdrawal effects or rebound insomnia after stopping. Real-world use can differ, particularly after long use or alongside other sedatives, so plan the stop with your prescriber (FDA label 14.2).

Does suvorexant cause weight gain?

Weight gain is not listed among the adverse reactions in the US label. The label does report a small dose-related rise in cholesterol (+2 mg/dL at 20 mg vs −4 mg/dL on placebo over 4 weeks). We found no independent study showing weight gain (FDA label 6.1).

Can I get Belsomra in the UK?

It is not licensed in the UK and does not appear in the UK electronic Medicines Compendium. The orexin antagonist appraised by NICE for NHS use is daridorexant, and NICE recommends it only when CBT-I has not worked, is not available or is unsuitable (EMC, NICE TA922).

Is suvorexant better than zolpidem?

No trial has compared them directly. In the Lancet 2022 network meta-analysis, zolpidem caused more side-effect dropouts than placebo, while suvorexant was grouped with drugs that can work short-term but have poor tolerability or missing long-term data. Neither was among the best-profile options (De Crescenzo et al., 2022).

Sources and funding notes

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

Have a question — or want us to cover something?

Ask about anything on this page, or request the next deep dive: an ingredient, a supplement, or a health concern. We use published research, evidence syntheses, and regulatory guidance, with clear source links.

We store your topic, message, optional email, and this page so we can manage and reply to the request. Do not include diagnoses, medications, or other sensitive medical information. See our Privacy Policy.