Why there is no dosage chart
A dose is established by trials that give different amounts to different groups and measure what happens, for a stated condition. That has been done for CBD in exactly one setting: severe childhood epilepsy, where a purified prescription product is dosed by body weight and titrated under medical supervision at levels far above anything sold in a shop.
For everything else — sleep, anxiety, aches, general wellbeing — no dose has been established, because the trials that would establish one have not been run. The charts on retail sites, matching body weight to milligrams by severity, are not simplified science. There is nothing underneath them.
Take it with food
Swallowed CBD is poorly absorbed and what does get absorbed is inconsistent, and the biggest lever on that is not the dose at all. In a phase I trial published in CNS Drugs, healthy volunteers took purified CBD fasted and again after a high-fat meal. The meal raised peak blood concentration roughly fourteen-fold and total exposure roughly four-fold.
Read that next to a dosage chart. The difference between taking the same capsule on an empty stomach and taking it after a meal is larger than the difference between most of the steps on the chart. Anyone comparing two products, or two weeks, without holding food constant is measuring their lunch.
How much of what you swallow arrives
A systematic review of CBD pharmacokinetics in humans, published in Frontiers in Pharmacology, pulled together what is known across routes. The pattern is consistent: oral bioavailability is low — a fraction of the dose, with wide variation between people — because the liver processes most of it before it reaches circulation.
| Route | What changes |
|---|---|
| Swallowed: capsule, gummy, drink | Through the gut and the liver first. The least of the dose arrives and the timing is the slowest, typically thirty minutes to two hours. |
| Sublingual: oil held under the tongue | Part of the dose crosses under the tongue and skips the first pass through the liver. Faster, and the dropper makes the amount adjustable. |
| Inhaled | The fastest and the most of the dose, minutes rather than hours, and the shortest-lived. |
| Topical | Local. Not a route to anything systemic, whatever the milligrams on the tub. |
None of that says which route is better. It says a milligram is not a milligram: the same number on two different products can deliver quite different amounts to the blood.
What a serving actually contains
Most confusion about CBD dosage is arithmetic rather than pharmacology. The large number on the front of the bottle is the whole container, and the number that matters is what one serving delivers.
| What the label says | What to do with it |
|---|---|
| A total for the bottle | Divide by the number of servings. A 30 ml tincture is usually about 30 droppers, so a bottle total divided by thirty is one full dropper. |
| Milligrams per millilitre | The most useful form. Multiply by how much of the dropper you actually take. |
| A count of gummies | Divide the total by the count, and check whether a serving is one sweet or two. |
| Nothing per serving at all | Read it as a product that would rather not be compared, and compare it with one that would. |
Two products advertising the same headline number can deliver quite different amounts per serving, and a smaller bottle is sometimes the stronger one per dropper. Doing this arithmetic once, before buying, removes most of the mystery from the category.
More is not reliably stronger
The instinct with a supplement that does nothing is to take more of it. CBD is a poor candidate for that instinct. In the one trial that compared doses side by side for anxiety, 300 mg lowered anxiety against placebo while 150 mg and 600 mg did not — an inverted U rather than a line.
That has been shown for one outcome in one setting, and it should not be over-generalised. But it is the only direct dose comparison in this literature, and it points away from working upwards until something happens.
What the high-dose trials reported
The fullest safety picture comes from the epilepsy programme, at 10 to 20 mg per kilogram of body weight per day — for most adults, several hundred milligrams, far above any retail serving.
| Reported | Worth knowing because |
|---|---|
| Sleepiness and sedation | The most frequent finding, and dose-related. |
| Diarrhoea, reduced appetite | The usual reasons people stopped. |
| Raised liver enzymes | Seen particularly alongside valproate, and monitored with blood tests in that setting. |
| Raised levels of other medicines | CBD inhibits cytochrome P450 enzymes in the liver; the clobazam interaction is the best documented. |
The interaction point is the one that survives at any dose. If you take a prescribed medicine, the question of whether CBD changes its level is a question for your pharmacist, answerable for your specific list and not in general.
What a real dosing protocol looks like
It is worth seeing one, because the contrast with a retail dosage chart is the argument of this page in a single comparison. The prescription CBD product used in severe childhood epilepsy is dosed by body weight and raised in steps under supervision.
| A prescription protocol | A retail dosage chart |
|---|---|
| Dosed per kilogram of body weight, twice a day | Dosed by a weight band, once, with no reference to how often |
| Started low and raised in defined steps, a week apart | Arrives at a final number immediately |
| Liver enzymes checked with blood tests before and during | No monitoring of any kind |
| Interacting medicines identified first and doses adjusted | Not mentioned |
| A named condition with trial evidence behind the dose | "Mild, moderate or severe", undefined |
The point is not that a supplement should come with blood tests. It is that the chart borrows the appearance of the left-hand column without any of its content, and the appearance is what makes it persuasive.
How to work out your own answer
Since no dose is established, the only honest method is to run a careful experiment on yourself and be strict about it.
- Hold food constant. Same relationship to a meal every time, given the fourteen-fold difference above.
- Hold the product constant. Same brand, same batch if you can, same format. The labelling studies make cross-product comparison close to meaningless.
- Change one thing at a time, and give each step a fortnight rather than a night.
- Write it down as you go. Dose, time, food, and what you were measuring, recorded at the time rather than remembered afterwards.
- Do not assume upwards. If a step does nothing, the next step up is not guaranteed to do more.
Reading a label without being taken in
All of the above assumes the product contains what it says. A 2017 analysis in JAMA of 84 CBD extracts bought online found around seven in ten were inaccurately labelled, some substantially under, some substantially over, and some with undeclared THC.
Four things on a label carry more weight than the rest of the packaging: milligrams per serving rather than per bottle, the spectrum, a batch certificate matching the code on your container, and the second ingredient — because a blend cannot tell you which part did anything.
Questions people ask
How many milligrams of CBD should I take?
No dose has been established for any over-the-counter purpose, so any specific number here would be invented. The only dose established in trials is for severe childhood epilepsy, by body weight, under medical supervision, at levels far above retail servings.
Does it matter whether I take CBD with food?
Considerably. In a phase I trial, a high-fat meal raised peak blood concentration around fourteen-fold and total exposure around four-fold compared with taking the same dose fasted.
Is a stronger bottle better?
Not reliably. In the one trial that compared doses directly for anxiety, the middle dose worked and the highest did not. Check milligrams per serving rather than per bottle before comparing two products at all.
How long should I stay at one dose before changing it?
Give it a fortnight, hold food and product constant, and write down what you were measuring at the time. Anything faster mostly measures the ordinary variation in whatever you are watching.
Sources
- Taylor L, et al. A Phase I, Randomized, Double-Blind, Placebo-Controlled, Single Ascending Dose, Multiple Dose, and Food Effect Trial of the Safety, Tolerability and Pharmacokinetics of Highly Purified Cannabidiol in Healthy Subjects. CNS Drugs. 2018;32(11):1053-1067. accessed Sep 18, 2026
- Millar SA, et al. A Systematic Review on the Pharmacokinetics of Cannabidiol in Humans. Front Pharmacol. 2018;9:1365. accessed Sep 18, 2026
- Linares IM, et al. Cannabidiol presents an inverted U-shaped dose-response curve in a simulated public speaking test. Braz J Psychiatry. 2019;41(1):9-14. accessed Sep 18, 2026
- Bonn-Miller MO, et al. Labeling Accuracy of Cannabidiol Extracts Sold Online. JAMA. 2017;318(17):1708-1709. accessed Sep 18, 2026
