CBD for sleep: 34 studies, 2 about insomnia

The most rigorous review of this question ever published gathered 34 studies. Two of them were about people who actually had insomnia — and one of those two was a single case report.

That number comes from a systematic review led by researchers at the Mayo Clinic, published in Cannabis and Cannabinoid Research in 2023. It is not an outlier finding buried in a hostile journal. It is the paper Google itself surfaces on the first page for this exact search, sitting between a national sleep charity and a shop’s product collection.

And there is a second number, easier to check, that most pages on this topic never print. Pick up almost any product sold as a CBD sleep aid, turn it over, and read the panel. You will usually find melatonin. Sometimes CBN. Frequently both.

So the honest position is this: the science has barely asked your question, and the shelf has quietly stopped asking it. Everything below is an attempt to show you both halves of that, with the numbers, so you can decide what to do with an incomplete answer rather than being handed a confident one.

What the best available review actually found

A systematic review is not one study. It is a structured sweep of everything published, scored for reliability, so that a weak result and a strong one do not carry equal weight. For this question, the 2023 review is the closest thing to a summary of the field that exists.

Thirty-four studies, and how few were about the right people

The team searched Ovid MEDLINE, Embase, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews and Scopus, from inception to December 2021. Thirty-four studies met the inclusion criteria.

Of those thirty-four, only two focused on patients with insomnia, and one of the two was a case report — a write-up of a single person. The other thirty-two measured sleep as a secondary outcome in people recruited for something else entirely: chronic pain, neurological conditions, substance use, general wellbeing.

That distinction is the whole ballgame. “Sleep improved in a group of people who were not having trouble sleeping” is a different sentence from “sleep improved in people who cannot sleep”, and only the second one answers the question you typed into a search box.

The split that changes the meaning of the result

Here is the part that almost never survives the journey from abstract to blog post. Of the 34 studies, 19 used a CBD-predominant preparation. Twenty-one used a near-equal ratio of CBD to delta-9 THC. The numbers overlap because some studies ran more than one arm.

Then look at which arms reached statistical significance:

• CBD-predominant arms: 4 of 7 studies that did hypothesis testing
• Near-equal CBD-to-THC arms: 12 of 16

The evidence base for “cannabinoids and sleep” is stronger than the evidence base for “CBD and sleep”, and the gap between the two is filled by THC. If you are buying a product with a 0.2 % or 0.3 % THC ceiling — which is what “CBD oil” means in every market where it is legally sold — the studies with the better hit rate were not studying your product.

Improvement without anything to compare it to

“All studies reported improvement in the insomnia symptoms of at least a portion of their participants.” That sentence sounds like a green light and is closer to a warning.

Sleep is one of the outcomes most susceptible to regression to the mean. People seek help on their worst nights, and their worst nights are, statistically, followed by better ones whatever happens next. Without a control group taking an identical-looking placebo, you cannot separate the compound from the calendar. Many of the 34 had no such group.

Questionnaires, not sleep labs

The review notes that several studies used non-validated subjective measures, and that most failed to include any objective measurement at all.

Objective here means polysomnography or at minimum actigraphy: electrodes or a wrist sensor recording when you actually fell asleep, how often you surfaced, how long you spent in each stage. Subjective means you rated your night on a scale in the morning.

Those two things diverge more than people expect. A compound that makes you feel calmer about a bad night can move a questionnaire score without moving a single minute of measured sleep latency. That is a real and interesting effect, but it is not the effect the product is sold for.

What the authors themselves concluded

Their conclusion is carefully hedged, and worth reading in the original rather than in summary. They write that the results suggest CBD alone or with equal quantities of THC may be beneficial, and then state plainly that research assessing effectiveness in a population of patients specifically with insomnia, using validated subjective and objective measures, is necessary before definitive inferences can be made.

That is the strongest available scientific statement on this question, in the authors’ own words. If a page tells you more than that, the extra confidence did not come from the literature.

The sleep aisle is not selling you a test of CBD

Now put the papers down and read a label. This half takes sixty seconds and is, in practice, more decisive than the first half.

Charlotte’s Web: 10 mg CBD, 3 mg melatonin

One of the largest and oldest hemp brands in the United States sells a product called Sleep CBD Gummies. The supplement facts panel lists 10 mg of cannabidiol and 3 mg of melatonin per two-gummy serving. The brand does not hide this — melatonin is in the product name on some of their pages — but it changes what the product is.

The panel also carries a warning that melatonin may cause drowsiness and that you should not drive after taking it. That warning is there because of the melatonin, not because of the CBD.

cbdMD: CBD, 20 mg CBN, 5 mg melatonin

Another major brand’s sleep gummies list CBD, 20 mg of CBN and 5 mg of melatonin per serving, depending on the concentration you pick. Three active ingredients, one of which is a hormone with an established sedative profile, and the CBD is the one on the front of the jar.

Cornbread Hemp: no melatonin, CBN instead

Not every brand does this, and it is only fair to say so. Cornbread Hemp markets its sleep line as melatonin-free and reaches for CBN instead. That is a genuinely different formulation choice, and it is the one to look for if you actually want to find out what cannabinoids do to your sleep.

It also proves the point rather than undermining it: even the brand that removes melatonin does not sell you plain CBD for sleep. It adds a different cannabinoid.

Why every formulator lands in the same place

This is not a conspiracy, it is product development. A company that sells a sleep product needs customers to notice something on night one. Melatonin reliably shortens the time it takes to fall asleep for a decent fraction of people, costs almost nothing, and is legal to sell in most markets as a supplement. The effect of CBD on sleep, per the review above, is uncertain enough that a company betting its refund rate on it would be taking a real risk.

So they hedge. The result is a category named after the ingredient with the marketing appeal, formulated around the ingredient with the reliable effect.

The one-minute check on any sleep product

Turn the jar around. Ignore the front of the label entirely — the front is a name, the back is a fact. On the supplement facts or ingredients panel, look for any of these:

Melatonin · CBN · GABA · 5-HTP · valerian · passionflower · chamomile · L-theanine · magnesium · lemon balm · hops

If any of them appear, the product is not a clean test of CBD, and if it works for you, you have learned nothing about cannabidiol. That is not a reason to avoid it. It is a reason to know what you bought.

Melatonin has an authorised claim, and CBD has none

The regulatory record makes the asymmetry official, and it is public and searchable.

What melatonin is allowed to say in Europe

Under Commission Regulation (EU) No 432/2012, melatonin carries an authorised health claim: it contributes to the reduction of time taken to fall asleep, and the claim may be used where the product provides at least 1 mg per portion. A second authorised claim covers the easing of jet lag at 0.5 mg.

Those are on the EU Register of nutrition and health claims, which anyone can search. They exist because a dossier was submitted, EFSA assessed the evidence, and the Commission voted.

What CBD is allowed to say

Nothing. There is not a single authorised health claim for cannabidiol in the EU register — not for sleep, not for anything. Under Regulation (EC) No 1924/2006, an unauthorised claim on a food is simply not permitted, whether or not it is true.

The only cannabidiol product in Europe permitted to make medical statements is Epidyolex, an EMA-approved prescription medicine, and its authorisation covers specific seizure conditions in named syndromes. It is not sold as a supplement and its file has nothing to do with sleep.

So why does the jar say “Sleep”?

Because a product name is not a health claim in the legal sense, and because enforcement of 1924/2006 against small consumer goods is patchy. “Sleep Gummies” as a name occupies a grey zone that “helps you sleep” as a sentence does not.

Once you know that, brand names stop reading as information. A product line called Sleep tells you which shelf it is on, not what it does.

What the asymmetry tells you about the formulation

Here is the useful inference. A company that adds 3 mg of melatonin is buying itself the right to a claim it can defend, on an ingredient with a regulatory file behind it. That is a rational thing to do — and it means the compound doing the regulated work is the one that is not in the product name.

CBN, the cannabinoid the aisle switched to

CBN turns up constantly in sleep formulations now, usually presented as the “sleepy cannabinoid”. Its story is worth a minute because it is more complicated than the marketing.

Where CBN comes from

CBN is not something the plant makes in quantity. It is what THC becomes as it ages, oxidises and degrades. Old, badly stored cannabis is high in CBN precisely because it has been sitting around. Commercially it is now produced deliberately, but its origin explains a lot about what follows.

The sedation claim and where it came from

The reputation traces back in large part to small mid-twentieth-century work and to the observation that old cannabis felt more sedating than fresh cannabis. The second observation has an obvious alternative explanation: old cannabis has less THC in it, and the experience of a weaker high is not the same thing as sedation.

Modern controlled evidence for CBN as a standalone sleep agent is thin — thinner than for CBD, which is saying something. It is a plausible candidate that has not been properly tested.

What CBN means for a drug test

This matters more than the sedation question for a lot of readers. CBN is a THC metabolite in origin, and some immunoassay screens are not perfectly specific. A product delivering 20 mg of CBN per serving is a larger cannabinoid load than most CBD oils, and the interaction between that and a workplace screen has not been characterised the way delta-9 has.

If you are screened at work, a high-CBN sleep gummy is a worse risk than a CBD oil, not a better one.

Reading CBN on a certificate

On a certificate of analysis, CBN appears as its own line in the cannabinoid profile. A CBD flower or oil with a notable CBN figure is telling you about its age and storage. A product with CBN added on purpose will say so on the panel, in milligrams, alongside the CBD.

If CBD does help you sleep, what is plausibly happening

Skepticism about the evidence is not the same as saying nothing happens. Plenty of people report a real change. Here are the mechanisms that would explain it without requiring CBD to be a sedative.

Being anxious before bed is not the same condition as insomnia

The strongest human evidence for CBD sits in anxiety-adjacent settings — public speaking models, small trials in anxious populations — not in sleep. If what keeps you awake is a racing mind rather than a broken sleep architecture, a compound acting on anxiety could plausibly change your night without acting on sleep at all.

That distinction shows up in the review’s structure too: most of the 34 studies measured sleep in people recruited for a different problem, and improving that problem is a perfectly good route to a better night.

Pain that wakes you is not insomnia either

Same logic. Several of the reviewed studies were in chronic pain populations. If discomfort is what surfaces you at 3 a.m., anything that changes the discomfort changes the sleep, and the sleep finding is downstream.

The gap between what studies used and what jars contain

This one is stark and rarely stated. Clinical work on CBD in humans has typically used quantities per day measured in the hundreds of milligrams. A sleep gummy delivers ten to fifty.

The gap is an order of magnitude, and it cuts in an uncomfortable direction: the amounts with any research behind them are not the amounts in consumer products, and the amounts in consumer products have essentially no research behind them. This site does not tell anyone what quantity to use — that is a conversation with a pharmacist or doctor — but the size of the gap is a fact worth carrying around.

Expectation, which is not an insult

If you spend money on something, take it deliberately at a fixed hour, and lie down expecting to sleep, you have built a bedtime ritual with a strong cue attached. Ritual and cue are among the better-supported things in sleep hygiene.

Calling that a placebo response makes it sound fake. It is not fake — the sleep is real sleep. It is just not evidence about a molecule.

Form and timing change what actually reaches you

If you are going to try anything, the delivery route matters more than most product pages admit, and it is settled science rather than speculation.

Swallowed: roughly 6 %

Anything you swallow — a gummy, a capsule, a brownie — goes through the liver before it reaches your bloodstream. First-pass metabolism destroys most of it. Oral bioavailability for CBD is commonly put at around 6 %, meaning most of the milligrams on the label never circulate.

Onset is slow with it: typically 60 to 120 minutes. A gummy taken as you switch the light off is arriving in the middle of the night, not at the start of it.

Held under the tongue: 13 to 19 %

Sublingual absorption bypasses part of that first pass, and reported figures land in the 13–19 % range. It requires actually holding the oil under the tongue for a minute or two rather than swallowing it immediately, which is the step most people skip.

Onset is faster, roughly 15 to 45 minutes.

Inhaled: about a third

Vaping or smoking puts a much larger fraction into circulation — on the order of a third — and does it within minutes. It is also the route with the clearest respiratory downside, and for a nightly habit that trade is worth thinking about carefully.

The high-fat meal multiplier

Taking oral CBD with a high-fat meal has been shown to increase absorption several-fold — figures in the region of four to five times are reported. It is the single largest variable in oral use and almost nothing on a jar mentions it.

Practically: the same gummy after dinner and the same gummy on an empty stomach are not the same event.

What that timeline means at bedtime

Put the two facts together. A swallowed product with slow onset, taken at lights-out, peaks somewhere in the small hours. If your problem is falling asleep, the timing is wrong. If your problem is waking at 3 a.m., the timing is accidentally right — which may explain part of why reports on this are so inconsistent.

What the label owes you, and what it hides

Total CBD versus CBD

A certificate may report CBD and CBDA separately. CBDA is the acid form the living plant makes, and heat converts it to CBD at a known ratio. The industry standard conversion is:

Total CBD = CBD + (CBDA × 0.877)

If a product quotes the total and you are comparing against one that quotes only the CBD line, you are not comparing like with like.

Per gummy or per bottle

“1500 mg” on the front of a jar is almost always the whole container. Divide by the count. A jar of 60 gummies at 1500 mg is 25 mg each, and the big number exists because big numbers sell.

Melatonin inside a proprietary blend

Some panels list a proprietary blend with a total weight and no breakdown. That is legal in the United States and it means you cannot tell how much melatonin you are getting. If sleep is the purpose and the blend is opaque, the product has failed the only test that matters here.

What a certificate of analysis adds

A batch-level certificate from an independent lab gives you the cannabinoid profile — CBD, CBDA, THC, THCA, CBN, CBG — plus contaminant panels. It will not list melatonin, because melatonin is not a cannabinoid and does not appear on a cannabinoid assay. For that you still need the supplement facts panel. You need both documents, and they answer different questions.

Interactions worth knowing before a nightly habit

The grapefruit rule

CBD inhibits several cytochrome P450 enzymes, notably CYP3A4 and CYP2C19, which between them handle a large share of common medicines. The practical shorthand doctors use: if your medication carries a grapefruit warning, the same pathway is involved, and CBD deserves the same caution.

This is not a fringe concern. It is documented in the EMA’s assessment file for Epidyolex, where interactions with clobazam and valproate are described in detail.

Sedatives stacked on sedatives

If a product contains melatonin and you are already taking something for sleep, you are combining two things that both act on the same outcome. Adding a second sedating ingredient on top of a prescribed one is exactly the situation where a pharmacist is the right person to ask, and a website is the wrong one.

Alcohol

A nightcap and a sleep gummy is a common combination and a poor one. Alcohol shortens sleep latency and then fragments the second half of the night. Stacking sedation on top of that does not fix the fragmentation.

When the answer is a pharmacist

Pregnancy, breastfeeding, any regular prescription, any liver condition, and anything you are giving a child. Also anyone screened at work. In each of those, the correct move is a professional who can see your full medication list, and no page on the internet can substitute for that.

How to run an honest test on yourself

If you want to know whether this does anything for you specifically, you can find out. It requires a little discipline and about a fortnight.

One variable at a time

Buy the plainest product you can find: CBD, carrier oil, nothing else. No melatonin, no CBN, no botanical blend. If the label lists five actives, the experiment is over before it starts.

Two weeks, not two nights

Two good nights prove nothing — you have two good nights most fortnights anyway. Give it long enough that normal variation averages out, keep everything else stable, and change nothing else during the window.

Record the clock, not the mood

Write down the time you got into bed and your best estimate of the time you fell asleep, plus the number of times you remember waking. A phone or watch that estimates sleep is imperfect but at least consistent night to night.

The reason to record numbers rather than impressions is the same reason the review criticised non-validated subjective measures: impressions move for reasons that have nothing to do with what you took.

What a negative result looks like

Decide in advance what would count as “no”. If after fourteen nights your average time to fall asleep has not moved, that is an answer, and it is worth as much as a yes. Most product experiments fail this step because no failure condition was ever defined, so the trial can only ever confirm.

The short version

What is established

Melatonin has an authorised European claim for shortening the time taken to fall asleep at 1 mg per portion. CBD has no authorised health claim anywhere in the EU register. Most commercial CBD sleep products contain melatonin, CBN, or both. Swallowed CBD is poorly absorbed and slow.

What is plausible but unproven

That CBD changes sleep indirectly, through anxiety or discomfort, in some people. The Mayo-led review found signals consistent with that, in populations recruited for other reasons, using measures the authors themselves flagged as inadequate.

What is marketing

The word “Sleep” on the front of a jar whose sedating ingredient is listed on the back. The 1500 mg that turns out to be 25. And any page that answers this question with more confidence than the people who read all thirty-four studies were willing to use.