CBD for Kids: What Parents Should Know About Evidence and Safety

If your child has seizures, autism, ADHD, anxiety or trouble sleeping, you have probably seen CBD described as a gentle, natural option. Here is what the research shows, and how strong it is. We do not give doses or recommend products.

At a glance

Epidiolex: the one approved CBD medicine for children

Epidiolex is a purified CBD oral solution, available only by prescription. Its prescribing information (revised May 2026) lists one use: seizures linked to Lennox-Gastaut syndrome (LGS), Dravet syndrome or tuberous sclerosis complex (TSC) in patients 1 year of age and older; safety and effectiveness under 1 year have not been established (DailyMed 2026). The FDA says that apart from Epidiolex and two medicines containing synthetic THC (Marinol and Syndros), no cannabis-derived product has been approved as safe and effective for any patient group, "whether pediatric or adult"; the same page also lists Cesamet, whose ingredient nabilone is a synthetic compound similar to THC, as FDA-approved (FDA 2024).

The label describes four randomized, double-blind, placebo-controlled trials in people whose seizures other medicines (which nearly all kept taking) had not controlled (DailyMed 2026):

• LGS (two trials, ages 2 to 55): over 14 weeks, drop seizures fell by a median of 44% versus 22% on placebo in one trial, and by 37% and 42% (two dose groups) versus 17% in the other.
• Dravet syndrome (120 patients, ages 2 to 18): convulsive seizures fell by a median of 39% versus 13%. Four of 60 children (7%) on Epidiolex had no convulsive seizures in the last 12 weeks, versus none on placebo.
• TSC (224 patients, ages 1 to 65): seizures fell by a median of 43% versus 20% over 16 weeks.

These results apply only to Epidiolex at the doses tested in the trials (set and adjusted by the child's doctor; this is not dosing advice). See Dravet syndrome, Lennox-Gastaut syndrome, tuberous sclerosis complex.

Strength of evidence: high for these three conditions. Randomized, double-blind, placebo-controlled trials of 14 to 16 weeks, reviewed by the FDA; they say nothing about other conditions or products.

What the label warns about

• Liver enzymes: in the LGS and Dravet trials, the liver enzyme ALT rose above 3 times the upper limit of normal in 13% of patients on Epidiolex versus 1% on placebo; the rate was 30% in those also taking valproate and clobazam, 21% with valproate alone and 3% with neither. Liver blood tests are done before starting, at 1, 3 and 6 months, and regularly after that (DailyMed 2026). See CBD and valproate.
• Sleepiness: 32% on Epidiolex versus 11% on placebo, and 46% of those also taking clobazam versus 16% of those not taking it (DailyMed 2026).
• Clobazam: Epidiolex raises clobazam's active breakdown product about 3-fold; the label tells doctors to consider a lower clobazam dose if side effects appear (DailyMed 2026). See CBD and clobazam and epilepsy medicines.

Strength of evidence: high. Placebo-controlled trials and drug-level studies in the FDA label.

Store-bought CBD: not checked for children

The FDA warns parents looking for treatments that "the use of untested drugs can have unpredictable and unintended consequences" (FDA 2024). Lab tests show why labels are not a reliable guide:

• Of 84 CBD products bought online in 2016, 31% were accurately labeled; 43% contained more CBD than stated and 26% less. THC was found in 21%, and the researchers wrote that it may be enough to cause intoxication or impairment, "especially among children" (Bonn-Miller 2017).
• Of 56 CBD gummy products, 70% had CBD more than 10% off the label; 39% contained delta-9 THC, and 9 had more than 0.4 mg of THC per gummy (Johnson 2025).

Strength of evidence: moderate. The lab analyses agree, but each tested a limited set of products at one point in time. See how to read a CBD label.

What research shows for autism, ADHD, anxiety and sleep

Autism ("CBD for kids on the spectrum")

• Israel: 150 young people aged 5 to 21 took a whole-plant extract, purified CBD plus THC (both 20 parts CBD to 1 part THC), or placebo for 12 weeks. One main measure showed no difference; on the other, 49% on the whole-plant extract were rated much or very much improved in disruptive behavior, versus 21% on placebo. Sleepiness and poor appetite were more common on the cannabinoids. The authors called the evidence for benefit "mixed and insufficient" (Aran 2021).
• United States: 39 boys aged 7 to 14 with severe behavior problems took Epidiolex and placebo for 8 weeks each. Both phases brought improvement on two of the three main measures, with no clear difference between CBD and placebo and a strong placebo effect; doctors' blinded impressions favored CBD in about two of three boys (Trauner 2026).
• Australia: in 29 children aged 5 to 12, a CBD oil with terpenes did no better than placebo on the main measure, social responsiveness, over 12 weeks; some secondary measures, including anxiety, improved (Parrella 2026).
• Brazil: in 60 children aged 5 to 11, a CBD-rich cannabis extract was linked to better social interaction, anxiety and agitation scores than placebo after 12 weeks (Silva Junior 2024).

A 2024 review of randomized trials in young people found modest effects, most consistent in autism, but too little evidence to broadly recommend cannabinoids (Köck 2024).

Strength of evidence: low. Four short trials (29 to 150 children) with different products, some containing THC, mixed results on their main measures and no long-term safety data. See also: autism.

ADHD

We found no randomized trial of CBD in children with ADHD. The only cannabinoid ADHD trial we found, in 30 adults, showed no significant difference from placebo on its main measure (Cooper 2017). It used Sativex (nabiximols), a mouth spray with roughly equal amounts of THC and CBD that is prescribed in some countries, such as England for multiple sclerosis, but is not FDA-approved (Javid 2024; FDA 2024). A 2026 review found too little data on ADHD to pool (Wilson 2026). See CBD and ADHD medicines.

Strength of evidence: no studies in children yet; one small adult trial of a THC-and-CBD prescription spray. See also: ADHD.

Anxiety

We found no placebo-controlled trial of CBD in children with anxiety disorders. In an Australian study, 31 people aged 12 to 25 whose anxiety had not improved with therapy or antidepressants added CBD for 12 weeks. Anxiety scores fell about 43%, but 25 of the 31 reported side effects such as tiredness and low mood, and there was no placebo group (Berger 2022). Two autism trials with placebo groups reported lower anxiety as a secondary result (Parrella 2026; Silva Junior 2024). The 2026 review (participants' median age 33) found no significant effect on anxiety (Wilson 2026).

Strength of evidence: very low for children. One small uncontrolled study plus secondary results in two small autism trials. See also: anxiety.

Sleep

In the Israeli autism trial, the CBD-rich treatments (which also contained some THC) were no better than placebo on any part of a children's sleep questionnaire, including bedtime resistance, time to fall asleep and sleep duration (Schnapp 2022). The Australian autism trial also measured sleep, but its summary does not report that result (Parrella 2026). The 2026 review found longer sleep with cannabinoids in insomnia trials (low-certainty evidence; participants' median age 33) (Wilson 2026). We found no randomized trial of CBD for sleep in other groups of children.

Strength of evidence: very low for children. See also: insomnia and CBD vs. melatonin.

THC edibles and accidental poisoning

• Poison center reports of children under 6 eating cannabis edibles rose from 207 in 2017 to 3,054 in 2021; 22.7% of the 7,043 children were admitted to the hospital (Tweet 2023).
• From January 2021 through May 2022, poison centers received 10,448 cases involving only THC edibles, and 91% of the unintentional exposures involved children and teens. The FDA warned that some edibles copy the packaging of popular cereals and candies (FDA 2022).
• The CDC found that cannabis-involved emergency visits among children aged 10 and under rose sharply during the pandemic, peaking at 71.5 a week in summer 2022 (CDC 2023).
• In 2023–2024, 11.3% of children 5 and under who ate edibles with hemp-derived cannabinoids such as delta-8 THC were admitted to critical care, versus 8.58% after delta-9 THC edibles; no deaths were reported (Weigel 2026). More: what is delta-8 THC?

Strength of evidence: national poison center and emergency surveillance. It shows clear trends and that cases can be serious, but not every case is reported.

What the law says

• Federal: until December 11, 2026, unless Congress delays the change again, federal law defines hemp by its delta-9 THC content (no more than 0.3% on a dry weight basis) (Pub. L. 115-334). From November 12, 2026, products with cannabinoids the plant cannot produce naturally are no longer hemp. Starting December 11, 2026, unless Congress delays it again, the rest takes effect, including the exclusion of cannabinoids synthesized or manufactured outside the plant, the switch to total THC (THC plus THCA), and a limit of 0.4 mg total THC per container for final products. The law's definition of a hemp-derived cannabinoid product excludes FDA-approved drugs such as Epidiolex (Pub. L. 119-37; Pub. L. 119-103). As of late October 2026, some members of Congress had introduced bills to delay the change by two more years or to repeal it; none had passed (CRS 2026). More: the federal hemp THC ban.
• States: Florida bars selling products that contain hemp extract and are meant to be swallowed or inhaled to anyone under 21 (Fla. Stat. 581.217). Iowa bars selling or giving a consumable hemp product to anyone under 21, and people under 21 may not buy, possess or consume one (Iowa Code ch. 204). Other states differ: state rules, Florida, Iowa.

Source type: enacted federal and state statutes (final law); the federal changes take effect in two steps and could change only if Congress acts. Not legal advice.

When to talk to your child's doctor

• Before trying any CBD product, talk with your child's doctor or pharmacist, especially if your child takes clobazam, valproate, an ADHD medicine or anything that causes sleepiness. See our drug interaction tool.
• If your child takes Epidiolex, keep liver tests on schedule. The label says to call the doctor right away for signs of liver problems such as yellow skin or eyes, dark urine, nausea or vomiting, and not to stop it without the doctor, because stopping suddenly can bring on more seizures (DailyMed 2026). Do not swap it for a store-bought product.
• If your child may have eaten a THC product, or a CBD product that might contain THC, contact a poison center or emergency services right away; do not wait for symptoms (FDA 2022). The FDA says the child should see a doctor or emergency department, especially if acting unusual or sick (FDA 2024). Bring the package.

Frequently asked questions

References