If something went wrong, here is where it goes.
Care first, then the record. This page routes a cannabinoid safety concern to the people who can act on it — emergency services, the product's manufacturer, and the authority responsible for such products where you live. It also sets out our corrections path, for when something we published is wrong.
If you or someone else is in immediate danger — difficulty breathing, chest pain, fainting or near-fainting, seizure, severe confusion, loss of consciousness, or any symptom that is getting rapidly worse — contact your local emergency services now, or your national poison-control / poisons-information service. If a clinician or pharmacist is reachable, they come first too.
CBG Atlas is not a medical service and cannot help in an emergency. No one is standing by here, nothing on this page reaches anyone able to treat you, and no part of this site should ever delay a call for care. Get help first. Everything below still works afterwards.
Report it to the people who can act on it.
An adverse-event report is only worth writing if it reaches somewhere it can change something — a batch that gets pulled, a label that gets corrected, a signal that accumulates across many reports until an authority acts. Two destinations do that. Send it to both; they do not talk to each other automatically.
The manufacturer, using the label.
The contact details are on the pack or the product page — a company name, an address, usually an email or a customer-contact form. Quote the batch or lot number printed on the pack: it is what lets them trace the batch, check their own testing records, and find out whether anyone else has reported the same thing. Keep the packaging and any remaining product; do not throw it away before you have reported.
Your national regulator, on the record.
Most countries run an adverse-event reporting system that members of the public can use directly, operated by the national medicines regulator, the food-safety or food-supplement authority, or a poisons centre — which one depends on how the product is classified where you live. Search for your national medicines or food-safety regulator's adverse-event or side-effect reporting service. We deliberately do not print agency names, links or phone numbers here: they differ by country and change over time, and a stale one on a safety page is worse than none.
A clinician, if you were unwell.
A clinician or pharmacist who knows your history can assess what happened, and in many systems can file a report on your behalf that carries clinical detail yours cannot. Tell them exactly what you took, including the lot number. This is care, not paperwork — it comes before either of the steps above.
We do not operate an adverse-event reporting system, and we cannot file a report for you. CBG Atlas is an editorial reference: no pharmacovigilance function, no clinical staff, no duty or ability to forward anything to an authority. Telling us would not put your experience on any official record — only the manufacturer and your regulator can do that, which is why they are the whole of the advice above. We would rather say this plainly than take a report we cannot act on.
Eight things that make a report usable.
Whatever form or inbox you end up using, these are the details that turn an account into something that can be acted on. Write them out once and reuse the same text for the manufacturer and for your regulator. Answer what you can and skip what you can't — a partial report is worth sending. Never delay getting medical care in order to write it.
What happened
Describe the event in your own words — what you noticed, how it started, how long it lasted, and whether it has resolved.
When it happened
The date, and the approximate time of day.
The product, and its lot number
Brand and product name, the cannabinoid(s) named on the label, the format (oil, capsule, gummy, vape, topical), and the batch or lot number printed on the pack. The lot number is the single most useful line in the whole report — it is how a manufacturer traces a batch. Keep the packaging.
Amount and timing, if known
How much was taken, how it was taken, and how long before the event. An estimate is genuinely useful — do not leave this blank because you are unsure.
What else was involved
What you were doing at the time, and anything else taken that day — including prescription medicines, other supplements, alcohol and caffeine.
Whether you sought care
Did you contact a clinician, pharmacist, poison-control service or emergency service? If so, what were you told?
Who you have already told
Whether you have already reported this to the manufacturer, to a regulator, or to a clinician — and any reference or case number you were given. It stops the same event being counted twice, and it links the records.
How you can be contacted
An address or number the manufacturer or the authority can follow up on. Many reporting systems accept a report without your name — check the wording on the form you are using before you decide.
You are about to hand health information to a company or to a public authority. Send only what the report needs: what you took, what happened, and how to reach you. You do not need to supply a full medical history or any identity document, and you should not send scans or photographs of them. Regulators publish their own privacy terms — read them on the form, and check whether the system accepts an anonymous or pharmacist-mediated report if you would rather not be named. Manufacturers are not bound by the same rules a regulator is.
Something on this site is wrong.
This one is ours. We are not the right destination for an adverse event, but we are the only possible destination for an error in what we published — and we mean it as an invitation. The platform's claim to authority rests on being correctable in public: every claim carries a grade, every grade traces to a gated registry entry, and a change to either is recorded in a tamper-evident audit chain rather than quietly swapped in.
A statement that outruns its evidence.
A sentence that reads as stronger than the grade beside it, or that drops a caveat the underlying study insisted on. Over-statement is the failure mode we care most about.
A citation that does not say what we say it says.
A mis-attributed finding, a study we have summarised wrongly, a paper that has since been retracted or corrected, or a link that resolves to the wrong record.
Something newer, or something missing.
Evidence published since our last registry snapshot, or a gap we have not admitted. Send the PMID or DOI — we do not add a citation we cannot verify against PubMed.
Where this site came into it.
If something you read here influenced what you did, the page address and the exact sentence matter to us. That is a content error too, and it is how we find our own mistakes.
We are not printing a contact address until it is routed to a real, monitored inbox. An address that silently discards mail is worse than none — you would believe you had been heard. Write your correction out now if you have one; note the page address, the exact sentence, and the PMID or DOI, and send it the moment an address appears here. In the meantime, what we already admit we do not know is set out in the gap-map, and every underlying study is browsable in the Research Library.
The cardiovascular note.
It belongs on this page because it is the most safety-relevant thing our registry holds — and because it is a good example of how little we are willing to claim.
CBG may influence blood pressure via α2-adrenoceptor activity; the direction and magnitude of any human effect are uncharacterized. This is graded Hypothesized — mechanistic plausibility only. The receptor activity is measured in the laboratory; no study has measured an actual blood-pressure effect of CBG in a person. That is not reassurance and it is not a warning: it is an admission that we do not know. If you take blood-pressure medication or have a cardiovascular condition, consult a clinician first.
CBG Atlas is a scientific reference. Nothing here diagnoses, treats, cures or prevents any disease, and nothing here is a substitute for a clinician who knows your history. CBG's evidence base is overwhelmingly preclinical — see the gap-map for what we do and do not have evidence for, and the grading standard for what each grade permits us to say. Peregrine Biopharma funds the platform and holds no editorial authority.
What we don't know — and why that matters here.
Most of this page is about what to do after something goes wrong. This part is about what the record does not contain. Our registry holds no study of CBG in pregnancy, in breastfeeding, in children, over the long term, or alongside prescription medicines. That is not a gap we are working around — it is the finding, and it is the reason for every default below.
The registry holds no study of CBG in pregnancy or breastfeeding. There is no evidence of harm here, and no evidence of safety either — there is nothing. Where a compound is untested in pregnancy, the responsible default is to avoid it and to raise it with your midwife, obstetrician or doctor, who can weigh your situation in a way a reference page cannot.
No study in our registry examines CBG in anyone under 18. Children are not small adults — dose, metabolism and development all differ — so adult findings, of which there are very few, cannot be scaled down. Do not give CBG to a child without guidance from a paediatric clinician.
Every human study in the registry is short. The one human trial of CBG's effects measured a single dose, on a single occasion, in 34 people; the registry's pharmacokinetic work is in mice, not people. Nothing tells us what daily use over months or years does. If you intend to take CBG regularly, that is a decision made without evidence, and worth making with a clinician who knows your history.
Two published human studies that are not registry entries are set out further down this page — a single-dose tolerability trial and a pharmacokinetic one. Neither closes this gap. The first covers a single day; the second measures exposure, not effect. See acute tolerability and the chronic limit.
The registry holds no study of CBG taken alongside any prescription medicine. We therefore cannot tell you a CBG interaction is unlikely, because we have not seen it tested. If you take prescription medication — particularly for blood pressure, the heart, epilepsy, blood thinning, or anything with a narrow safe range — speak to a pharmacist or prescriber before starting CBG. A pharmacist can check your specific list; we cannot.
This connects to the one mechanism we have measured: CBG is a potent α2-adrenoceptor agonist in vitro (Cascio et al., PMID 20002104), and α2 agonism can lower blood pressure. Whether that happens in a person is uncharacterised — see the cardiovascular note above.
A reference that only publishes what it knows will read as more confident than the evidence deserves. These four gaps are among the most consequential things we could tell you about CBG, and every one of them is an absence. We would rather say so plainly than leave you to assume the silence means safety.
What one dose showed — and what it cannot tell you.
Healthy adults have been given single oral doses of CBG under trial conditions and tolerated them. Nobody has given CBG to a person every day for weeks or months and measured what happened. Both of those are printed here, together, always — because the first one on its own reads like a clearance it is not.
In a single-ascending-dose trial, twelve healthy adults received single oral doses of CBG isolate in MCT oil, ascending through 25, 50, 100 and 200 mg, with a placebo randomised into the dose sequence. No drug-related adverse events were reported, and liver function tests did not exceed the upper limit of normal after either of the two highest doses. Blood levels rose in dose order, but varied considerably from person to person. Subjective effects were minimal, with three exceptions: participants rated themselves significantly less “Jittery” and less “Active” at 50 mg, significantly less “Calm” at 100 mg — a change in the opposite direction to the one CBG is popularly sold for — and reported more appetite at 200 mg. The authors' own conclusion is the sentence worth carrying: CBG was well tolerated with no robust pharmacodynamic effects, and chronic dosing studies are needed. (Wolinsky et al., 2026, J Pharmacol Exp Ther 2026:104984, PMID 42575778, trial registration NCT05324982.)
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. Twelve people, on single occasions — a tolerability signal, not a safety guarantee. Disclosed interest: one co-author is affiliated with Charlotte's Web, a cannabinoid product company.
Chronic human safety is not established. No published study has given CBG to people repeatedly over weeks or months and measured what happened. The result above covers single occasions; it tells you nothing about the fiftieth dose, which is why the trial's own authors say chronic dosing studies are needed.
That space is not empty, either. In a 90-day study in rats, CBG given daily by orogastric tube at 0.66 and 1.33 mg/kg/day moved several markers the wrong way: total oxidative stress rose significantly, malondialdehyde and carbonylated proteins were elevated, the white cell count was disrupted, and ALT activity, creatinine and ionised calcium were altered — alongside regressive (hepatotoxic) changes in the liver. CBD at 0.66 mg/kg/day did the opposite, improving redox status; the authors attribute the divergence to the extra dimethyloctadienyl group CBG carries. (Polanska et al., 2023, Toxicology 488:153460, PMID 36796712.)
Read that carefully, because it is easy to over-read. It is rats, at low doses, and an animal finding does not establish that CBG harms a person — species differ, and this has never been looked for in a human. It is on this page for one reason: it is precisely why nobody should call repeated human use of CBG well tolerated on the strength of a single-dose trial. Normal liver tests on one day are not a normal liver at ninety days, and we do not have the second measurement in people. If you intend to take CBG daily, that is a decision made without human evidence, and worth making with a clinician who knows your history.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here — and an animal toxicology finding could not ground a human safety grade in any case.
Because apart, each one misleads. “No drug-related adverse events in healthy adults” on its own reads as a clearance CBG has not been given. “Regressive liver changes at 90 days” on its own reads as a warning about people that rat data cannot support. The honest position is the pair: a single oral dose was tolerated in a small trial, and what repeated use does in a human is simply unmeasured. Neither sentence is publishable here without the other — that is a standing rule, not a decision about this page.
The same dose is not the same exposure.
This is the most practical finding in the published human record, and it is about absorption rather than benefit. What you eat alongside an oral CBG dose changes how much of it reaches your bloodstream — by enough that two identical labels can behave like two different products.
In a double-blind, double-crossover pilot trial, participants took a single 25 mg oral dose of CBG under different conditions, with concentrations measured by tandem mass spectrometry. A high-fat meal significantly raised total exposure, peak concentration and terminal half-life compared with a low-fat meal — and the difference was wide enough that the threshold for bioequivalence between the two conditions was not met. Dietary fat affected exposure more than the emulsified delivery vehicle did. (Story et al., 2024, Cannabis Cannabinoid Res 9(6):1543-1555, PMID 38574248.)
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. The number of participants is not stated in the published abstract, and we do not print a figure we cannot verify — read it as a pilot study, on that basis alone.
It means the same number of milligrams can put different amounts of CBG into your blood depending on what you ate, and that dietary fat is the larger lever — larger, in this trial, than the delivery format being tested. If you are comparing two products, or wondering why one occasion felt unlike another, the meal is a variable worth holding constant before you draw any conclusion about the product.
It does not mean more is better. This is pharmacokinetics: it measures where a compound goes and how long it stays, not what it does to you. Our registry holds no study linking a CBG blood level to any outcome in a person, so a higher level is not a stronger benefit — it is simply a higher level. Taking more, or taking it with a fatty meal to “get more out of it”, is not a strategy this evidence supports.
CBDV is not CBG. Its safety record is its own.
Everything above this section is about CBG. Everything in this section is about CBDV (cannabidivarin), a different compound with a different — and much larger — human record. It is here because a reader with a safety question about a cannabinoid arrives on this page, not because the two records are interchangeable. They are not, and nothing below should be read as describing CBG.
CBDV is the propyl homolog of CBD — the same cannabidiol scaffold, two carbons shorter. Three things follow, and all three are easy to get wrong:
CBD's record does not transfer. CBD has an approved prescription medicine. CBDV has no approved use in any jurisdiction, for any indication. They were tested separately and they got different results; a shared scaffold is shared chemistry, not shared evidence.
CBG's record does not transfer either — in either direction. The α2-adrenoceptor blood-pressure note above is a CBG finding and says nothing about CBDV. The food-effect finding above is a CBG pharmacokinetic study and does not extend to CBDV. Equally, none of the CBDV signals below tells you anything about CBG.
CBDV has actually been given to people, repeatedly, in registered trials — which is what makes this section possible and is unusual for a minor cannabinoid. What those trials found is set out below and in full on the CBDV pillar.
In a randomised, double-blind, placebo-controlled Phase 2 trial, 162 adults with inadequately controlled focal seizures took CBDV or placebo as add-on therapy, titrated to 800 mg twice daily. Start with the result, because the rest is unreadable without it: the trial did not work. Seizure frequency fell 40.5% on CBDV and 37.7% on placebo (p = 0.648). The primary efficacy endpoint was not met.
And CBDV produced substantially more adverse effects than placebo. Treatment-emergent adverse events were reported by 72.8% of the CBDV group against 48.1% on placebo; the three most common were diarrhoea, nausea and somnolence. Serious adverse events were uncommon in both arms — 3.7% on CBDV against 1.2% on placebo. The trial reported little or no effect on vital signs, physical examination or ECG. (Brodie et al., 2021, Cannabis Cannabinoid Res 6(6):528-536, PMID 33998885, trial registration NCT02365610.)
The two halves are one finding, and we will not print either alone. An adverse-event profile on its own invites the obvious assumption — that this is what a working drug costs. It is not. This is what CBDV cost the people who took it while not separating from placebo. Read together, the pair says something a safety page should say plainly: on the strongest evidence anyone has produced, there was a price and there was no measured benefit to set against it.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. Disclosed interest: this is a sponsor study — the authors are affiliated with GW Research Ltd and Greenwich Biosciences, Inc.
This is the finding on this page most worth carrying away. In the Phase 2 trial, alanine or aspartate aminotransferase rose above three times the upper limit of normal in three participants taking CBDV — two of whom discontinued as a result — against one participant on placebo. None met the criteria for a potential Hy's Law case, the pattern regulators treat as a marker of serious drug-induced liver injury. (Brodie et al., 2021, PMID 33998885.)
A second, entirely separate study found the same organ system moving. In an open-label Phase 1 study in five girls with Rett syndrome and drug-resistant epilepsy, a transient transaminase rise — smaller, staying below one and a half times the upper limit of normal — prompted a dose reduction. (Hurley et al., 2022, Epilepsia 63(7):1736-1747, PMID 35364618.)
Two research groups, two populations, two designs, one signal. That is not a diagnosis and it is not a claim that CBDV damages livers — the numbers are small, most participants were unaffected, and none of it has been characterised over the long term. It is the reason CBDV is treated here as a research-stage compound rather than a benign ingredient, and the reason anyone with liver disease, or taking a medicine that is watched for liver effects, should raise a CBDV product with a clinician before taking it rather than after.
Verified against their PubMed records and cited in prose. Neither paper is yet a graded entry in the CBG Atlas registry, so neither carries an evidence grade here.
In that same open-label study — five participants, no control arm, no blinding, no randomisation, dosed up to 10 mg/kg/day — the EEG worsened in two of the five. Spike-wave activity doubled in one participant and rose more than 500-fold in another, with a rhythm the investigators considered likely ictal.
The paper does not report this consistently, and we are not going to resolve it for you. Its abstract summary line records no significant change in EEG; its Results report the worsening described above. We print both readings and pick neither: we cannot tell from the published record which the authors intended, and quietly choosing one — particularly the reassuring one — would hide a discrepancy a reader is entitled to see. If this matters to a decision you are making, read the paper itself. (Hurley et al., 2022, PMID 35364618.)
Hold the design in mind in both directions. Five participants with no control arm cannot attribute a change to the drug — not the worsening, and not the seizure reduction the study also reported. This is a tolerability and pharmacokinetic study being asked a question it was not built to answer.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. Disclosed interest: the declared conflicts of interest are “none”, but GW Pharmaceuticals supplied the study drug and set the dose ceiling — a relationship worth naming even where the declaration does not.
In laboratory assays, CBDV inhibited CYP2C19 and CYP1A1 — two of the enzyme systems the liver uses to clear many medicines. Read what those studies actually were before you weigh the finding: both are papers about cannabidiol, in which CBDV appears as one of several structurally related comparator compounds used to work out which part of the CBD molecule does the inhibiting. In the CYP2C19 work CBDV was slightly less potent than CBD. That qualitative comparison is the whole of what this page will say about strength — the potency figures are researcher-tier and are not reproduced on a consumer surface. (Jiang et al., 2013, PMID 23318708; Yamaori et al., 2013, PMID 23811569.)
CYP1A1 — not CYP1A2. They are different enzymes, and the substitution is a real error that circulates in secondary write-ups. No published CBDV data exists for CYP1A2, CYP2B6 or CYP2C9, so this page asserts nothing about them — and that absence is not a reassurance either. It is the same “no primary evidence” state as the gaps set out further up this page.
Both halves of this matter, and the second is doing the most work. An enzyme inhibited in a test tube is a signal worth acting cautiously on; it is not a demonstrated interaction in a person. No CBDV drug-interaction study in people has been published, so the practical size of this — whether it moves the blood level of anything you actually take — is unknown. Where it matters most is where the stakes are highest: anyone on antiseizure polypharmacy, or on any medicine with a narrow safe range. A pharmacist can check your specific list against CYP2C19; we cannot, and a reference page should not try.
Verified against their PubMed records and cited in prose. Neither paper is yet a graded entry in the CBG Atlas registry, so neither carries an evidence grade here. Both are in vitro enzyme studies in which CBDV was a comparator rather than the subject — read them as a reason for caution, not as a measured interaction.
Nothing on this page supports stopping, reducing or replacing a prescribed antiseizure medicine — or any other prescribed medicine — with CBDV. CBDV is not an approved medicine in any jurisdiction, for any indication. In the trial that tested it for seizures it was given on top of existing antiseizure medication, as an add-on, and it did not beat placebo. Changes to epilepsy treatment belong with the clinician managing it; stopping an antiseizure medicine without one is dangerous in its own right. This is educational information, not medical advice.
A CBDV food-effect figure. A widely repeated claim holds that a high-fat meal multiplies CBDV exposure several times over. No published paper reports it — it traces to a trial document, and we do not print a number we cannot check against a published source. The food-effect section above is about CBG (Story 2024) and does not extend to CBDV, which has no published food-effect study of its own.
Two registered trials, cited as though they produced results. They did not: both were terminated, with a handful of participants between them, and neither supports any statement about safety or effect. Where a termination reason is not verifiable, none is given here — a plausible-sounding cause attached to an unverified record is a fabricated fact, not a helpful gloss. Their status is set out on the CBDV pillar.
Every CBDV paper cited on this page — published, off-registry, ungraded
A Phase 2 Randomized Controlled Trial of the Efficacy and Safety of Cannabidivarin as Add-on Therapy in Participants with Inadequately Controlled Focal Seizures
Phase 2 RCT · in people · primary endpoint not met
Efficacy and safety of cannabidivarin treatment of epilepsy in girls with Rett syndrome: A phase 1 clinical trial
Phase 1 · open-label, non-randomised, no control · n = 5
Cannabidiol is a potent inhibitor of the catalytic activity of cytochrome P450 2C19
In vitro enzyme assay · CBDV as comparator
Structural requirements for potent direct inhibition of human cytochrome P450 1A1 by cannabidiol: role of pentylresorcinol moiety
In vitro enzyme assay · CBDV as comparator
Daily Use of a Broad-Spectrum Cannabidiol Supplement Produces Detectable Concentrations of Cannabinoids in Urine Prohibited by the World Anti-Doping Agency: An Effect Amplified by Exercise
Regulatory & testing status · not an evidence claim
The four sources above, plus the anti-doping paper cited in the anti-doping section, which measured CBG and CBDV in the same study. Every one is real, published and checkable, and not one is a graded entry in the CBG Atlas registry. They are cited in prose for exactly that reason, and their identifiers are not carried in any study feed. The line under each title is its study design, which is what limits it — it is not an evidence grade, and no grade on this platform is assigned outside the gated publish.
CBN is sold for sleep. The trials missed their primary endpoints.
Everything above is about CBG, except the CBDV section, which is about CBDV. This section is about CBN (cannabinol) — a third compound, with its own record, and the most heavily marketed cannabinoid this platform covers. It is here because the single most consumer-relevant finding in the whole cannabinoid safety literature right now is about a CBN product combination sold on shelves today. Nothing below describes CBG or CBDV.
CBN is not made from CBG, and it is not a plant product in the way the others are. There is no CBN synthase. The plant commits CBGA to three enzymes that end at THC, CBD and CBC — none of them makes CBN. CBN forms afterwards, by oxidation of Δ9-THC and THCA under air, heat, light and time. It accumulates in aged and poorly stored material, which makes its level closer to a freshness measurement than a cultivar one. This is the single most misunderstood fact about CBN and it is settled chemistry rather than a finding.
Coming from THC transfers nothing. THC's effects, THC's legal status and THC's clinical record belong to THC. A molecule is not its precursor — and being a breakdown product of an intoxicant is not evidence that CBN is intoxicating, nor evidence that it is mild. Read CBN's own record, below and on the CBN pillar.
CBG's and CBDV's records do not transfer either. The α2-adrenoceptor blood-pressure note above is a CBG finding and says nothing about CBN. The food-effect finding above is a CBG study and does not extend to CBN. The liver-enzyme signal in the section above is CBDV's, from different studies in different people.
This is the most consumer-relevant finding on this page, and it is about a product pairing rather than a compound on its own. CBN + melatonin is a mass-market retail combination — the two are routinely sold together in a single sleep capsule or gummy. A 2025 study looked at what happens when you take them together.
It found that CBN potently inhibits CYP1A2, the main liver enzyme that clears melatonin, and that giving CBN alongside melatonin to mice produced a roughly four-fold increase in melatonin exposure in the blood. The same group had previously reported CBN inhibiting the same enzyme's handling of caffeine. In plain terms: the mechanism suggests the melatonin dose printed on a combination product may not be the melatonin exposure it delivers. (Anderson et al., 2025, Basic Clin Pharmacol Toxicol 136(1):e14120, PMID 39722474, doi:10.1111/bcpt.14120.)
Now read the limit, because it is doing as much work as the finding. This is laboratory enzyme assays and mice. It is not a demonstrated interaction in a person. No CBN drug-interaction study in people has been published, so whether this happens in a human — and how large it would be — is genuinely unknown. That unknown is not a reassurance. It is on this page precisely because the combination is already being sold and nobody has measured it in the people buying it. If you take a CBN + melatonin product, or take CBN alongside any medicine with a narrow safe range, a pharmacist can check your specific list against CYP1A2. We cannot, and a reference page should not try.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here — and a preclinical finding could not ground a human interaction grade in any case. The inhibition strength is a researcher-tier value and is not reproduced on this consumer surface.
Start with the endpoints, because the rest is unreadable without them. Both CBN sleep trials that pre-specified a primary endpoint missed it. In a randomised placebo-controlled trial of 293 adults taking 20 mg nightly for seven nights, the primary endpoint was sleep quality and it was not significant (p = .082); two secondary measures were — fewer nighttime awakenings (p = .025) and lower overall sleep disturbance (p = .023) — while sleep onset latency, wake after sleep onset and daytime fatigue showed no difference from placebo at all. (Bonn-Miller et al., 2024, Exp Clin Psychopharmacol 32(3):277-284, PMID 37796540.)
In the second, 20 adults with physician-diagnosed insomnia disorder took 30 mg, 300 mg and placebo on separate nights under overnight polysomnography. The primary endpoint was wake after sleep onset, and CBN did not significantly change it at either dose (300 mg: p = 0.29; 30 mg: p = 0.50). At 300 mg the trial did report more stage-2 non-REM sleep, better subjective sleep quality, shorter sleep onset latency and fewer EEG arousals — all of them ranked below the primary in the trial's registered outcome hierarchy. (Lavender et al., 2026, J Sleep Res 35(4):e70284, PMID 41698831, trial registration NCT05344170.)
The tolerability picture, which belongs in the same box. The polysomnography trial reported 247 mild-to-moderate adverse events across its arms. The seven-night trial reported no impact on daytime fatigue at 20 mg — worth knowing, and not the same thing as a clean bill. Neither trial ran long enough to say anything about repeated nightly use over months: the longest CBN exposure in any published controlled trial is measured in weeks.
Both halves are one finding, and we will not print either alone. Positive secondary measures on their own read as a working sleep aid; they are not. They are what CBN produced while missing the question each trial set itself. And the mirror error matters just as much: a missed primary endpoint is a finding, not an absence. These trials were run properly and they were answered. Anyone telling you there is “no evidence” on CBN and sleep is as wrong as anyone telling you it is proven.
Verified against their PubMed records and cited in prose. Neither paper is yet a graded entry in the CBG Atlas registry, so neither carries an evidence grade here. Disclosed interest: all nine authors of the seven-night trial are employed by cannabinoid companies — eight at Canopy Growth Corporation and one at Wana Brands. The polysomnography trial is academic, and is a single night per arm in twenty people.
This is the most practical thing on this page, and it applies whatever you make of the trials above. Every CBN dose that produced any measured signal is four to sixty times the dose typically sold. A 2021 review of the CBN sleep claim put the retail figure on the record and called for trials using “dosages significantly higher than those found in currently available cannabis products marketed for sleep (typically ≤5 mg)”. Its overall conclusion, verbatim, was that “there is insufficient published evidence to support sleep-related claims”. (Corroon, 2021, Cannabis Cannabinoid Res 6(5):366-371, PMID 34468204.)
The trials since have done exactly that and nothing else. The lowest dose that moved anything is 20 mg — four times that benchmark. The dose behind the polysomnography trial's positive measures is 300 mg — sixty times it, and in that same trial 30 mg moved essentially nothing. So the published record has very little to say about what a ≤5 mg product does, and what it does say runs the wrong way.
Read that as an absence, not a refutation. Nobody has shown that a retail dose fails, because nobody has properly tested one — and the answer to a dose gap is not to take more. Taking a multiple of a label dose to reach a trial dose is self-experimentation with a compound whose repeated-use safety in people is uncharacterised, and it is not a strategy this evidence supports.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. It is a narrative review, single-authored — its value is that it searched specifically for the trials the marketing implies exist, and reported what it found.
No published study has measured residual THC in products sold as CBN. We are not going to invent one. The question is a reasonable one to ask — CBN is produced from THC, and a compound made by degrading THC is an obvious place for THC to be left behind — but a reasonable question is not a finding, and this platform does not publish the difference between them as though it were closed.
What does exist is adjacent evidence about labelling, in a different product category, and it is offered here on that basis only. A survey of 500 commercially available cannabis oils in Argentina analysed by mass spectrometry included 17 products carrying a “THC-free” label claim. Of those 17, 76.5% — thirteen — tested positive for Δ9-THC. The authors report wide variability in cannabinoid content, purity and labelling across the whole sample. (Fernández et al., 2023, Forensic Sci Int 349:111762, PMID 37392612, doi:10.1016/j.forsciint.2023.111762.)
That is a fact about cannabis oils in one market in 2023. It is not a fact about CBN products, and it must never be quoted as one. What it supports is a general and modest point: a cannabinoid label is a claim, not a test result, and a “free of X” claim was wrong for most of the products that made it in that survey. The practical response is the same one this page gives everywhere — read the product's certificate of analysis, check it names the batch in your hand, and if a specific cannabinoid matters to you for testing, employment or medical reasons, treat an untested label as unverified.
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here — and it could not carry a CBN grade in any case, because it is not a CBN study. It is cited as an explicit analogy about labelling practice, and its limits (one country, one year, cannabis oils rather than isolate sleep products, and only 17 products in the THC-free subgroup) are the reason it is an analogy rather than a measurement.
CBN is not a substitute for a prescribed hypnotic, or for any other prescribed medicine, and nothing on this page supports stopping or replacing one. CBN is not an approved medicine in any jurisdiction, for any indication. The two trials that pre-specified what they were testing did not meet it. Persistent insomnia is worth taking to a clinician in its own right — it is often a symptom of something treatable, and a reference page cannot assess that. This is educational information, not medical advice, and nothing here diagnoses, treats, cures or prevents any disease.
Every paper cited in this section — published, off-registry, ungraded
Cannabinol and Sleep: Separating Fact from Fiction
Narrative review · published 2021, BEFORE all three CBN sleep trials
A double-blind, randomized, placebo-controlled study of the safety and effects of CBN with and without CBD on sleep quality
Randomised placebo-controlled trial · human · primary endpoint NOT met
Cannabinol for Acute Treatment of Insomnia Disorder in a Randomized Placebo-Controlled Crossover Trial
Randomised placebo-controlled crossover trial · human · polysomnography · primary endpoint NOT met
Preclinical Evidence for a Drug-Drug Interaction Between Cannabinol and Melatonin
⚠ Preclinical · liver-enzyme assays and MICE — not an interaction study in people
An assessment of qualitative and quantitative cannabinoids analysis in selected commercially available cannabis oils in Argentina
⚠ NOT A CBN STUDY — a label-accuracy survey of cannabis oils, cited only as an analogy
Five sources, every one real, published and checkable — and not one a graded entry in the CBG Atlas registry, which holds no CBN study at all. They are cited in prose for exactly that reason, and their identifiers are not carried in any study feed. The last of the five is not a CBN paper: it is a label-accuracy survey of cannabis oils, cited once as an analogy and labelled as one everywhere it appears. The line under each title is its study design, which is what limits it — it is not an evidence grade, and no grade on this platform is assigned outside the gated publish. The efficacy record in full, including the two large trials not repeated here, is on the CBN pillar.
The anti-doping note.
This one is not a health matter — it is a regulatory and testing one, and it is on the safety page because for one group of readers it has a concrete, foreseeable consequence: a failed test. It carries no evidence grade, and it says nothing about whether CBG or CBDV does anything for anyone. It covers both compounds, because one study measured both — and each figure below is labelled with the molecule it belongs to.
CBG is prohibited by the World Anti-Doping Agency — CBD is the only exempt cannabinoid. In a randomised trial of a broad-spectrum CBD product containing only trace CBG, CBG was detectable in urine after repeated daily use, and concentrations appeared higher after exercise. This is regulatory and testing status, not a health claim. Check the current WADA Prohibited List and your product's COA. (Gillham et al., PMID 40920736.)
Cited as an attributed regulatory and testing note. This paper is in intake review and is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here.
The concern is not resting on one paper. A separate ethics-approved human study administered commercial hemp products as single doses and analysed volunteers' urine afterwards, with the explicit subject of the risk of unintentional anti-doping rule violations from consuming hemp products — that is, from the cannabinoids other than CBD that such products contain. Two independent groups, different designs, the same practical warning: a product bought and used as a CBD product can carry minor cannabinoids, and CBD is the only one WADA exempts. (Mareck et al., 2023, Drug Test Anal 15(1):27-41, PMID 35633098; the author group includes researchers at the German Sport University Cologne.)
Verified against its PubMed record and cited in prose. This paper is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. We print no urine concentration and no volunteer count from it: the published abstract itemises neither, and a figure we cannot verify would make this note sound more precise than the evidence behind it. If a threshold matters to your eligibility, it comes from your anti-doping organisation, not from us.
The note at the top of this section is about CBG. The same trial also measured CBDV, and found it at a higher detection rate. After ten weeks of daily use of a broad-spectrum CBD product that contained only trace CBDV, CBDV was detected in 68% of pre-exercise urine samples and 84% of post-exercise samples, with measured concentrations rising from before to after exercise. Nothing was detectable at baseline. CBDV is prohibited by the World Anti-Doping Agency too — CBD is the exempt cannabinoid, and the exemption does not extend to its propyl homolog. The authors' own conclusion is that athletes should avoid broad-spectrum CBD products. (Gillham et al., 2025, PMID 40920736 — the same paper cited above, read here for a different compound.)
This is detection and regulatory status, not a health claim: it says nothing about whether CBDV does anything for anyone, and nothing about whether it is safe. The trials that asked those questions are set out under CBDV safety, above. What it does say is practical and specific: a product bought, labelled and used as a CBD product put a prohibited cannabinoid into most users' urine, and exercise raised the reading.
Verified against its PubMed record and cited in prose. This paper is in intake review and is not yet a graded entry in the CBG Atlas registry, so it carries no evidence grade here. The CBG figures in the note above and the CBDV figures here come from the same study and are not interchangeable — CBG and CBDV are different molecules and were detected at different rates.
We do not publish the Prohibited List, and we are not an anti-doping authority: the list is revised, and a stale copy here would be exactly the kind of confident error this platform exists to avoid. Check the current list with the World Anti-Doping Agency or your national anti-doping organisation, and ask them before using any cannabinoid product. Read the product's certificate of analysis too — a broad-spectrum or full-spectrum extract can contain minor cannabinoids the front label never names. Nothing here is medical, legal or eligibility advice.