CBD Oil, Capsules and Gummies: What the Label and the Evidence Actually Tell You

Search for CBD and the shelf sorts itself into three camps: a dropper bottle, a pot of capsules and a bag of gummies. They all say "CBD", they all print different numbers, and each makes a decent case for being the sensible way in. Before you pick a camp, it helps to know what a format actually changes and what the label can tell you.

Here's the gist. The format changes how you take CBD and how easy it is to measure a serving. It doesn't reliably change how much CBD ends up in your blood in any way a label can promise: human studies show that food, formulation and the individual matter at least as much as oil versus capsule, and none of the studies below gives one trustworthy absorption percentage per format. What you can check is how many milligrams are in a serving, what else is in the extract, and whether a batch report backs the label up.

Same ingredient, different packaging

• Format: Oil (drops); What the label can usefully tell you: Total mg per bottle, bottle volume, drops per bottle, carrier oil; What to check: Convert to mg per drop and per serving (next section); Worth knowing: Drop size varies with bottle, dropper and oil thickness, so a per-drop figure is only as good as the drop count behind it
• Format: Capsules and softgels; What the label can usefully tell you: mg per capsule, capsule count, fill (oil or powder), shell material; What to check: Gelatine versus plant shell if that matters to you; whether the fill is an oil; Worth knowing: Easy to measure, but easy-to-measure isn't the same as better absorbed
• Format: Gummies; What the label can usefully tell you: mg per gummy, number per pack, other ingredients such as vitamins or sugars; What to check: Whether the stated mg is per gummy or per pack; Worth knowing: We found no head-to-head human study of gummies against oils or capsules
• Format: Everything else (topicals, inhaled products); What the label can usefully tell you: Different jobs entirely; What to check: Don't compare mg across routes; Worth knowing: A skin cream isn't an oral dose, and inhaled CBD is a different exposure again

Whatever the shape, you're comparing amounts of the same compound, mixed with a carrier and perhaps other plant compounds, made by a manufacturer whose batch may or may not match its label.

From bottle to milligrams: the label arithmetic

Say a bottle reads: 10 ml, 10% CBD, 250 drops. Four steps, no spreadsheet required, and you can repeat them for any bottle.

1. A "10%" oil usually means 10 g of CBD per 100 ml, or 100 mg per ml. (That's the usual reading; check whether the label says weight by weight instead, which changes the figure slightly.)
2. 10 ml × 100 mg/ml = 1,000 mg in the bottle.
3. 1,000 mg ÷ 250 drops = 4 mg per drop.
4. Daily amount = mg per drop × drops per serving × servings per day. Use the serving printed on your own label.

The same method works for other formats. A 60-softgel pack with 1,000 mg in total is 1,000 ÷ 60 ≈ 16.7 mg per softgel. A pack of 10 gummies at 25 mg each is 250 mg, so 25 mg is the figure per gummy, not per pack.

Once you have a daily figure, it helps to know what it sits next to. Here are three published reference points, each for a specific population and product type. None is a limit for a product on a shelf, none is advice, and they aren't directly comparable with one another. To convert EFSA's figure, multiply 0.0275 mg by your body weight in kg (the 70 kg column is only an example):

• Reference point: EFSA provisional safe level, 9 February 2026; What it says: 0.0275 mg/kg body weight a day, for food supplement formulations with CBD purity of at least 98%, without nanoparticles, with a safe production process and genotoxicity excluded; For a 70 kg adult: about 1.9 mg a day (EFSA gives roughly 2 mg)
• Reference point: German Senate Commission on Food Safety (SKLM) review, 2025; What it says: Liver enzyme rises seen in healthy volunteers from 4.3 mg/kg a day for 3 to 4 weeks (the lowest dose where effects were seen; lower doses were not tested); For a 70 kg adult: about 300 mg a day
• Reference point: Florian and colleagues, 2025, JAMA Internal Medicine; What it says: 5 mg/kg a day of an oral CBD solution for 28 days: liver enzyme elevation above three times the upper limit of normal in 8 CBD participants (5.6 per cent) and none on placebo; For a 70 kg adult: about 350 mg a day

EFSA's announcement says the level applies solely to those high-purity formulations, that safety couldn't be established for people under 25, pregnant or breastfeeding women, or people taking medicines, and that data gaps remain on the liver and on the endocrine, nervous and reproductive systems. Our reading is that it isn't a statement about full-spectrum extracts. The SKLM review found no evidence substantiating health claims at or below its lowest-effect dose, and concluded that this dose could easily be reached, or exceeded, at the doses recommended for products on the market. SKLM's own example was 30 drops of an oil containing 20% CBD. How many milligrams that is depends on how many drops make a millilitre, which varies by bottle; the worked example above uses its label's own count of 25 drops per ml.

"Which absorbs best?" has no tidy number

Search results love a neat percentage ("20 to 35 per cent under the tongue, 6 per cent swallowed"). Human research doesn't actually supply one.

What the pharmacokinetic literature has measured

A systematic review of human CBD pharmacokinetics (Millar and colleagues, 2018, Frontiers in Pharmacology) went through 792 articles and found 24 with pharmacokinetic data. Absolute bioavailability, meaning the share of a dose reaching the blood compared with an injection, was reported only for smoking, at 31 per cent. No study in that review reported it for other routes. The review did find that peak concentration rises when CBD is taken with food and in lipid (fat-based) formulations, and that the half-life after regular oral use was 2 to 5 days. That long half-life is worth keeping in mind: a number of hours on a label describes how a product feels to take, not how long CBD stays in the body.

Under the tongue or swallowed?

The usual instruction is to hold oil under the tongue for a minute or so before swallowing. A small crossover study in eight healthy men (Johnson and colleagues, 2024, Cannabis and Cannabinoid Research) compared the same hemp oil, 69 mg of CBD per millilitre, taken as sublingual drops and as gelatine capsules. Blood levels were similar: peak concentrations of 28.0 versus 24.0 ng/mL, both peaking around four hours, with no statistically significant difference. The authors' reading was that drops are probably swallowed before much absorption through the mouth lining happens. The study is tiny, used one oil, and can't tell us whether a longer hold would change things. It does mean "sublingual absorbs far better" shouldn't be treated as settled.

Food matters a great deal

Two sources put numbers on the food effect, and neither involves a retail oil. The first is a study of eight adults with refractory epilepsy who qualified for Minnesota's cannabis programme and took 99 per cent pure CBD capsules (Birnbaum and colleagues, 2019, Epilepsia): the peak concentration averaged about 14 times higher and total exposure about 4 times higher after a high-fat meal of 840 to 860 calories than when fasting. The ranges were wide (the 90 per cent confidence interval for peak concentration ran from 7.5 to 31.9 times). The US prescribing information for Epidiolex reports, for healthy volunteers given 750 or 1,500 mg, a 5-fold higher peak and 4-fold higher exposure after a high-fat, high-calorie meal, and a 4-fold and 3-fold rise after a low-fat meal. It adds that alcohol raised the peak by 93 per cent and exposure by 63 per cent. FDA prescribing information, revised June 2025, section 12.3.

The European product information for the same medicine tells patients to take it consistently either with or without food, precisely because food changes the amount. For a retail product the practical point is not a recipe but a reminder: the same capsule on an empty stomach and after a big meal isn't the same dose.

Formulation can change the curve

One randomised, double-blind crossover study of 16 healthy volunteers (Knaub and colleagues, 2019, Molecules) gave 25 mg of CBD, fasted, in a proprietary self-emulsifying formulation or in the same hemp extract diluted in MCT oil. The self-emulsifying version produced a 4.4-fold higher peak and about 2.9-fold higher exposure over eight hours, with the peak arriving after about one hour rather than three. The study was funded by Vesifact AG, the company behind the self-emulsifying technology. It's evidence that formulation can matter, and equally that "water-soluble", "liposomal" or "nano" on a label is a claim about one specific formulation, not a property of a format. Gummies, oils and capsules each contain many different formulations.

What to do with all this

Treat oil, capsules and gummies as different ways of measuring and taking a serving, not as a ranking of strength. If a seller gives an absorption percentage for a format, ask which study it comes from, in whom, fasted or fed, and with what formulation.

Full-spectrum, broad-spectrum, isolate: does the bottle match the label?

These three words describe what stays in the extract, and they're used loosely:

• Full-spectrum usually means the extract keeps other plant compounds alongside CBD, such as minor cannabinoids, terpenes and a small amount of THC.
• Broad-spectrum usually means other plant compounds are kept but THC has been removed or reduced.
• Isolate is CBD purified to a single compound, usually sold as a crystalline powder or used as an ingredient.

Whether a particular bottle fits its label is a testable question. In a US study of 202 CBD products (Gidal and colleagues, 2024, Frontiers in Pharmacology), 26 per cent did not meet the definition of the product type claimed on the pack, and 74 per cent deviated from their labelled CBD potency by at least 10 per cent. The sample was American products and doesn't describe any brand you might be looking at. A European analysis of 186 CBD and related cannabinoid oils collected over about a decade (Binova and colleagues, 2026, Analytical and Bioanalytical Chemistry) found 86 of 181 did not contain the declared amount of CBD. Several contained THC isomers beyond the usual Δ9-THC, including Δ8-THC and Δ10-THC, and one product marketed as full-spectrum contained HHC. The authors estimated that the intake of Δ9- and Δ8-THC at recommended servings would exceed EFSA's acute reference dose of 1 microgram per kilogram of body weight in 39 per cent of the products. Those oils were gathered over many years, so they're a reason to want batch evidence, not a current failure rate.

A systematic review of drug testing (Ovat and colleagues, 2026, International Journal of Legal Medicine) points the same way: positive cannabinoid findings after CBD or hemp oil use mainly come from product content, such as undeclared or excess THC, mislabelling or contamination. Five of eight human studies it reviewed reported no result above test cut-offs, though some levels sat close to the detection limits. If you drive for work, take part in tested sport or are subject to workplace testing, "THC-free" on a label is a claim to check against a batch report.

The entourage effect: a hypothesis with a marketing career

The entourage effect is the idea that CBD works better alongside the other compounds of the plant than on its own. It's a respectable hypothesis, and it's also a sales line. Here is where each one stands, in brief:

• Source: Ben-Shabat and colleagues, 1998, European Journal of Pharmacology; What it was: Mouse and tissue study of the body's own cannabinoid system, where inactive fatty-acid glycerol esters boosted the endocannabinoid 2-AG; What it found: Coined the phrase. It wasn't about hemp; a 2023 scoping review (Christensen and colleagues, Biomedicines) notes the term was applied to cannabis later, as a hypothetical afterthought
• Source: Russo, 2011, British Journal of Pharmacology; What it was: Review making the case for plant terpenes as possible contributors; What it found: A proposal for future experiments, explicitly "if proven"
• Source: Santiago and colleagues, 2019; Finlay and colleagues, 2020; What it was: Laboratory tests of terpenes on human CB1 and CB2 receptors; What it found: No direct effect and no change to THC's signalling; no effect on binding or function, apart from a possible weak effect of β-caryophyllene at CB2. Both say this doesn't rule out effects through other targets
• Source: Pamplona and colleagues, 2018, Frontiers in Neurology; What it was: Meta-analysis of observational epilepsy studies; What it found: 71 per cent of patients on CBD-rich extracts reported improvement against 46 per cent on purified CBD, but people, products and doses differed (6.0 versus 25.3 mg/kg/day) and the authors said controlled studies are needed
• Source: Dahlgren and colleagues, 2022; What it was: Open-label study of 14 outpatients with anxiety taking a full-spectrum, high-CBD sublingual solution; What it found: Reported improvement, but with no placebo and no isolate comparison it can't speak to full-spectrum versus isolate

The 2023 scoping review summarised the field as in its infancy, with simple methods and contradictory results, and suggested that ordinary pharmacology (synergy, bioenhancement) may explain much of what is observed.

What does that mean for a label? "Full-spectrum" is a description of composition that a batch report can confirm. "Full-spectrum works better because of the entourage effect" is a hypothesis that hasn't been shown to apply to retail CBD products at the amounts they contain. A product can reasonably be chosen for its composition without that claim being treated as established.

Reading a batch report (COA)

A certificate of analysis (COA) reports what a laboratory measured in one sample. It becomes useful when the sample is traceable to the product in your hand.

• Match the batch. Product name, formulation and batch number on the report should match the pack, and the sample should be the finished product rather than only an ingredient.
• See what was tested. A cannabinoid panel and a contaminant panel answer different questions, and "not detected" means below a stated reporting limit, not necessarily zero.
• Read the units. A concentration per gram or millilitre and a total per container are different numbers; convert them the way you did for the label above.
• Check the laboratory. Accreditation covers defined activities. The UK body UKAS, for instance, says a schedule defining the scope is made public on its website; look at the equivalent register for the laboratory's own country, and whether the test you care about is in scope.
• Keep quality and benefit apart. A clean, well-documented batch tells you about composition. It doesn't show that the product works or that it suits your medicines.

What one supplier says about its formats

The details below come from Cibdol's public pages, read on 8 October 2026. They are the supplier's statements, not our laboratory tests, and not a recommendation.

Cibdol's pages list these formats for CBD: oil in 10 ml dropper bottles in six strengths from 5 to 40 per cent, softgels in the same six strengths, and mango and raspberry gummies. It also lists CBG and CBN oils and skincare. On its CBD oil 10% page the supplier states 1,000 mg CBD per 10 ml bottle, 250 drops and roughly 4 mg per drop, with a suggested use held under the tongue for 60 seconds. For the 10% softgels it states 60 softgels, about 16.7 mg each, and one softgel three times a day; the shells contain bovine gelatine, so they're not vegetarian. For gummies it states 10 per pack at 25 mg CBD each and vitamin B12.

Cibdol describes its products as "full-spectrum 2.0" extracts with measured CBN, CBG, CBC and CBDa, says every batch is analysed by an independent laboratory, and says certificates of analysis can be opened by batch number. A sample certificate on its CBD oil 10% page shows THC below 0.05 per cent. We haven't reviewed Cibdol's certificates beyond that sample, so treat these as the supplier's claims and check the batch report on your own pack, as with any seller. Product ranges, strengths and prices change. On 8 October 2026 the single-bottle prices on its pages were €45 for the 10% oil, €45 for the 10% softgels and €9.95 for a gummy pack (the page also lists lower per-bottle prices for multi-bottle purchases), which works out at €4.50 per 100 mg of CBD for the oil and softgels and about €3.98 per 100 mg for the gummies.

Medicines, pregnancy and the legal picture

Medicines. CBD affects liver enzymes that process many drugs. The Epidiolex prescribing information documents a roughly three-fold rise in an active clobazam metabolite and about a 2.5-fold rise in everolimus exposure in dedicated studies. In the epilepsy trials behind the prescribing information, most liver enzyme rises happened in people who were also taking valproate. If you take prescription medicines, particularly for epilepsy, transplant or blood thinning, ask a pharmacist before adding CBD in any format. Our guide to how supplements and medicines interact explains how to prepare for that conversation.

Pregnancy and breastfeeding. EFSA couldn't establish safety for these groups, and the European product information for CBD medicine advises against use in pregnancy unless the benefit clearly outweighs the risk and advises stopping breastfeeding during treatment.

EU and Dutch rules, read on 8 October 2026. EFSA's 9 February 2026 announcement says the European Commission considers that CBD qualifies as a novel food provided it meets the conditions of EU novel-food legislation, and that EFSA will assess each CBD novel food application on the data supplied. A 2025 review by the German Senate Commission on Food Safety said that no CBD food products had yet been authorised in the EU but that they remained on sale. We couldn't confirm the EU novel-food status of CBD products beyond that, so we make no statement about it. Check the Commission's Novel Food Status Catalogue before buying. In the Netherlands we found no official THC limit for CBD products. DRUGSinfo (undated) mentions 0.2%; undated law-firm commentary mentions 0.05%. Dutch sources differ, so check the current NVWA position. Laws change, they differ between countries, and "on sale in a shop" isn't the same as "authorised"; check current rules before you buy or travel.

Choosing a format: what to take away

• Choose a format for how you want to take it: drops if you like adjusting a serving, capsules or gummies if you want a fixed amount per piece.
• Don't choose it for an absorption percentage. Human studies show that food, formulation and the individual shift exposure a lot, and no study we found gives one per format.
• Compare labels in mg per serving and mg per day, using the arithmetic above. With or without food? For the prescription medicine, European guidance is to be consistent either way; for a retail product, the lesson is that a meal changes the dose.
• Treat "full-spectrum", "broad-spectrum" and "THC-free" as claims about composition, and ask for the batch report that backs them up.
• Treat the entourage effect as an interesting hypothesis, not a reason to pay more.
• Check medicines, pregnancy and driving before you start, whatever the format.

For more on reading labels in general, see our guide to how to read supplement labels.

Where to read next

On the legal side:

• Is CBD Legal in Europe? EU Law and the Dutch Position, for how EU novel food status, EFSA's 2 mg figure, THC numbers and the Dutch position fit together.

Written by Smart Supplements with AI assistance from the sources linked in the text. It has not been reviewed by a clinician.

Last updated: 12 October 2026

This article is for informational purposes only and is not intended as medical advice. Always consult a healthcare professional before starting any new supplement, especially if you take prescription medication.