Autism and CBD Oil: Research, Dosage and Warnings

What does the research say?

Six small studies comparing products with placebo enrolled between 29 and 150 children and teens; the studies lasted 8–12 weeks. The products used varied widely: pure CBD, extracts containing small amounts of THC, and full-spectrum extracts. In the two studies that tested pure CBD, there was no clear difference between CBD and placebo on the main scales measuring behavior problems; in one, children on placebo also improved clearly.

In a 150-person study with an extract containing one part THC to twenty parts CBD, clear improvement according to doctors' overall assessment was seen in 49% of those treated with the extract and 21% of those on placebo; however, one of the two main measures was negative. A manufacturer-funded, 8-week study of 61 children with an extract containing very little THC and made up mainly of CBDA, the acid form of CBD, also reported a positive result on overall illness severity; however, the analysis included only the 54 children who completed the study. The evidence for THC-containing products is also weak: the studies are small and short, and their results do not agree with each other. On the other hand, serious mental health problems such as psychosis and mania were reported in some young people with THC-containing products.

Studies without a comparison group, based on families' observations, reported improvements in meltdowns, sleep and anxiety; these accounts are valuable, but they are not enough to tell what caused the improvement. According to Israel's national records, only about a quarter of autistic children started on medical cannabis continued treatment for a year; in a study group followed in detail, the main reasons for stopping were side effects and insufficient effect. There are no controlled studies in adults (apart from single-dose brain imaging studies) or in epilepsy that occurs alongside autism.

Fragile X syndrome is an inherited condition caused by a change in the FMR1 gene; it is the most common inherited cause of intellectual disability and the most common single-gene cause of autism, and anxiety and behavior problems are also common. In this syndrome, laboratory-made pure CBD with no THC was tested as a gel applied to the skin. In a 12-week controlled study of 212 children (CONNECT-FX), the main measure, the social avoidance score, improved similarly in the CBD and placebo groups. In a subgroup analysis added afterward, which was not in the study's original plan, a small difference was seen in children in whom the gene is most likely silenced; an exploratory finding like this does not count as proof of effect on its own. A second controlled study of 257 people in these children (RECONNECT) ended in August 2025; as of October 2026, its results had not been published either in a peer-reviewed journal or in the study registry. Studies without a control group (20 children; 240 children in long-term extension) reported improvement, and the gel was generally well tolerated. The evidence is weak in this syndrome too; all the studies were funded by the companies that developed the product.

Some children and young people with intellectual disability (marked difficulty with learning and everyday living skills) have severe behavior problems such as outbursts of anger, aggression or self-harm; some of these children are also autistic. On this question, pure CBD has been compared with placebo only in a small 8-child pilot study. The aim of the study was not to measure the effect but to test whether such a study could be done: all the children completed the study, no serious side effects were seen, and the authors reported a sign of an effect in favor of CBD. However, the number of children was far too small to draw a conclusion, and according to a review in which the team also took part, all the children taking CBD were autistic, while only one of those taking placebo was. That is why there is no controlled study giving a separate result with CBD for intellectual disability without autism. The same team's large controlled study with 132 children ended in August 2025; as of October 2026, its results have not been published. A 20-person pilot study without a comparison group with nabilone, a synthetic THC-like drug, for aggression in adults with intellectual disability was also completed in August 2025; as of October 2026, its result has not been published. There is not enough research on this question yet; there are studies whose results are awaited.

Strength of evidence: ●●○○ Weak evidence. Finding: No effect found. What does this mean?

Ongoing research on this topic

Trials testing CBD for this topic on ClinicalTrials.gov: 5 under way (4 recruiting). A trial being under way does not show that CBD works.

NCT06526208: Phase 2, recruiting, target 108 participants
NCT07199218: Phase 2, recruiting, target 78 participants
NCT04517799: Phase 2, active, not recruiting, 42 participants

All trials on this topic and their history

Possible side effects and drug interactions

The most common side effects were sleepiness and reduced appetite: in one study, sleepiness occurred in 28% with the THC-containing extract, 23% with pure CBD and 8% with placebo. Vomiting, restlessness and irritability were also reported; in some children, behavior got worse. With high-dose pure CBD, liver enzymes also went up in some cases. With CBD gel applied to the skin in fragile X syndrome, the most common treatment-related side effect was pain or redness where it was applied.

CBD can slow the breakdown of some medicines in the body; so caution is needed when it is combined with antipsychotics, antidepressants and epilepsy medicines. In one study, all three children who had serious side effects were taking more than one medicine, and in two of them an antipsychotic had been stopped suddenly without supervision.

Interactions with medicines used for this condition

Interactions reported for these medicines in peer-reviewed publications and official prescribing information. A medicine that is not listed can still interact with CBD; tell your doctor about every medicine you take.

Autism: medicines used and their interactions with CBD
Medicine	Importance	Warning	Source
Clobazam	High importance	Clobazam: If you take clobazam, CBD can raise the medicine's active metabolite about 3-fold and can increase drowsiness. Do not start without talking to your neurologist.	FDA Epidiolex label §7.2, §12.3 (05/2026), EMA prescribing information §4.5, PMID 26114620
Valproate (valproic acid) (Depakin, Convulex)	High importance	Valproate and the liver: Liver enzyme elevations were more common when CBD was used together with valproate. Your doctor should monitor your liver tests.	DailyMed Epidiolex label 5.1, PMID 28782097
Topiramate, Rufinamide, Zonisamide, Eslicarbazepine	Caution	Other epilepsy medicines: With epilepsy medicines, CBD can change the blood levels of some of them. Your doctor may monitor your drug levels and side effects.	PMID 28782097, PMID 33541771, PMID 37593907, PMID 31211851, PMID 30791225
Risperidone (Risperdal, Rileptid, Neoris), Other antipsychotics (haloperidol, etc.) (Norodol)	Information	Antipsychotics: No published study has measured antipsychotic blood levels with CBD. In schizophrenia studies, CBD (600–1000 mg a day) added for 6 weeks to the antipsychotic being taken was generally well tolerated; in one study, drowsiness was more common in those taking CBD. Drowsiness may increase when they are taken together; tell your doctor.	PMID 29241357, PMID 29619533, FDA Epidiolex label §7.4 (05/2026)

To check interactions with the medicines you take: CBD drug interaction checker.

How was CBD used in studies?

In the studies, the dose was calculated as milligrams per kilogram of body weight per day. For pure CBD or products with a very low THC ratio, the target dose was 5–20 mg per kilogram a day (study dose; not a personal recommendation). For extracts containing one part THC to twenty parts CBD, the target was 7.5–10 mg CBD per kilogram a day (study dose; not a personal recommendation). In studies where the dose was adjusted for each person (most with extracts that also contained THC), the average doses actually reached stayed between 0.7 and 5.9 mg per kilogram a day (study dose; not a personal recommendation). In the two controlled studies with pure CBD, however, the dose was raised on the same schedule for everyone to 10 or 20 mg per kilogram a day (study dose; not a personal recommendation).

In the studies, the dose was adjusted slowly over a few weeks and split into 2–3 portions a day.

In the fragile X syndrome studies, pure CBD was applied to the skin as a gel: 250 mg a day for those weighing up to 35 kg and 500 mg for heavier children (study dose; not a personal recommendation); in the long-term extension, the dose was adjusted between 125 and 750 mg a day.

Study details (for health professionals)

There are 6 small placebo-controlled studies (n 29–150, 8–12 weeks), and the products vary widely: pure CBD, 20:1 and 9:1 CBD:THC extracts, and a full-spectrum extract with a THC ratio <0.3%. The two RCTs with pure CBD were not clearly different from placebo on the primary outcomes; positive findings were seen on secondary measures or in a manufacturer-supported study. For sleep, two RCTs showed no advantage over placebo; there are no controlled data in adults or in epilepsy occurring alongside ASD. The placebo effect is strong, and most uncontrolled studies rely on parent reports. The "No effect found" label is based on the evidence for pure CBD (on this site, the CBD label is given for pure CBD, and products that also contain THC are rated separately): in the two controlled studies of pure CBD in autism, the primary outcomes were negative (the fragile X study was not used as a basis for the label because it was done in a different condition); positive findings were seen with extracts that also contain more or less THC (20:1 and 9:1 CBD:THC extracts, and a full-spectrum extract with a THC ratio below 0.3%) and mostly on secondary measures; the only study that was positive on a primary measure was manufacturer-supported. In fragile X syndrome, the 212-child RCT with pure CBD gel applied to the skin (CONNECT-FX) was negative on the primary outcome; a difference was seen only in an exploratory subgroup analysis that was not in the study's original plan, and the results of the second, 257-person RCT in this subgroup (RECONNECT) had not been published as of October 2026, either in the registry or in a peer-reviewed journal. October 2026 search (PubMed, Cochrane, ClinicalTrials.gov): 7 controlled studies with cannabinoids in autism (6 treatment studies, one with results only in the registry; one single-dose brain imaging study in adults; one sub-analysis was not counted separately and is mentioned in the main study's row), 1 controlled study in fragile X syndrome, 1 small controlled study in intellectual disability and severe behavior problems (8 children; feasibility study) and 4 systematic reviews were found; all are in the table below. There is no Cochrane review assessing cannabinoids. There are three more controlled studies with unpublished results: 86 children with pure CBD in autism (CASCADE), 257 people in fragile X (RECONNECT) and 132 children with pure CBD in intellectual disability (Australia; ended in August 2025). Publications from the same Israeli group of children were combined in a single row in the table (Hacohen 2022). In an additional search in October 2026 (intellectual disability and severe behavior problems only; PubMed, Cochrane, ClinicalTrials.gov, Europe PMC), the only controlled study on this question to be included in the table was the 8-child pilot study (Efron 2021; all the children in the CBD group were autistic, so there is no separate result for intellectual disability without autism). The results of a 20-person uncontrolled pilot study with nabilone in adults with intellectual disability (ended in August 2025) and of single-patient crossover trials for behavior problems in rare genetic conditions (tuberous sclerosis, MPS III, fragile X) that began in the Netherlands in 2025 had not been published as of October 2026.

Evidence assessment: Pure CBD: Weak (in autism, only two controlled studies, 39 and 103 children, 8–12 weeks, one with results only in the registry; in both, the primary outcomes were negative; positive findings come from small studies without a control group). The studies of gel applied to the skin in fragile X syndrome were not used as a basis for this level because they were done in a different condition, and neither was the 8-child pilot study in intellectual disability, because it was not designed to measure the effect (feasibility study; the groups were not compared statistically). The results of CBD-rich products that also contain THC or have an unclear THC amount were assessed under the THC evidence

CBD alone

In the two controlled studies with pure CBD (Trauner 2026: 39 boys aged 7–14, crossover; NCT04745026: 103 children, results only in the registry), no advantage over placebo was seen on the primary outcomes for behavior problems; the single-dose brain imaging study in adults (Pretzsch 2019) did not measure clinical effect. Two small studies without a control group (23 and 20 children) reported improvement. In fragile X syndrome, pure CBD gel applied to the skin was no different from placebo on the primary outcome in a controlled study of 212 children (CONNECT-FX); the difference in a subgroup analysis that was not in the original plan is exploratory and has not yet been confirmed. In children with intellectual disability and severe behavior problems, pure CBD has been compared with placebo only in an 8-child feasibility study (Efron 2021; the authors reported a signal in favor of CBD, but the number of children is far too small for a conclusion, and all the children in the CBD group were autistic); the result of the large 132-child study had not been published as of October 2026.

Products containing THC

There are four controlled studies with CBD-rich extracts that also contain more or less THC or have an unclear THC amount, and their results are not consistent. In a 150-child study with a 20:1 CBD:THC extract (Aran 2021), only one of the two main measures, the doctor's overall assessment, was better than placebo (clear improvement in 49% versus 21%); the other main measure was negative. In a 60-child study with a 9:1 CBD:THC extract (Silva 2024), there was no difference on standard scales; only some symptoms improved on an interview form prepared by the authors. In a 34-child crossover study with a CBD oil reported to contain at most 1 mg THC per milliliter (Parrella 2026), the main measure was negative and some secondary measures were positive. In a manufacturer-funded study of 61 children with a full-spectrum extract containing very little THC and made up mainly of CBDA, the acid form of CBD (Keating 2026), the main measure (overall illness severity) was positive; the analysis included only the 54 children who completed the study. A meta-analysis pooling two of these studies found a small effect (Köck 2024); a 2026 meta-analysis pooling all cannabinoid products also saw a small reduction in autistic traits; this result rests on 2 studies done with products containing both THC and CBD, with low certainty of evidence (Wilson 2026). Studies and registries without a control group from Israel, Brazil, Turkey and the UK reported improvement; in Israel's national data, only about a quarter of children continued treatment for a year. Cases of psychosis and mania were reported in some young people with THC-containing products.

Evidence: Weak (four short RCTs in 34–150 children with CBD-rich extracts that also contain THC or have an unclear THC amount: one positive on the main measure but manufacturer-funded and analyzed only in completers, one positive on one of two main measures, one positive only on the authors' interview form, one negative on the main measure; low certainty in meta-analyses; uncontrolled studies are low-level evidence)

Dose range in studies

In the studies, target 5–20 mg/kg/day for pure CBD or low-THC products; 7.5–10 mg/kg/day CBD for THC-containing 20:1 extracts (THC ceiling 0.375–0.5 mg/kg/day); averages reached 0.7–5.9 mg/kg/day (in studies where the dose was adjusted for each person, most with extracts that also contained THC; in the two controlled studies with pure CBD, the dose was the same for everyone, 10 or 20 mg/kg/day); in fragile X, pure CBD gel applied to the skin at 250 or 500 mg a day depending on weight (125–750 mg in the extension)

Starting dose and titration

In the studies, the dose was started low (e.g. 1 mg/kg/day or 5 mg/kg/day, 1–6 drops a day) and raised at rates ranging from +1 mg/kg every other day to +5 mg/kg a week; reaching the target took 2 weeks to 2 months; the total dose was split into 2–3 a day; when stopping, it was tapered over 1–2 weeks

These are doses researchers gave under medical supervision; they are not a recommended dose for you. "mg/kg/day" means milligrams per kilogram of body weight per day.

Autism and CBD: studies reviewed
Study	Participants	Product	Dose and titration	Duration	Result	Source
Aran 2021 Double-blind, placebo-controlled RCT, 3 arms, prespecified crossover (efficacy from period 1 only)	150 ASD, aged 5–21 (mean 11.8)	(1) Whole-plant extract CBD:THC 20:1; (2) pure CBD + pure THC 20:1. 167 mg/ml CBD, 8.35 mg/ml THC	Started at 1 mg/kg/day CBD (+0.05 mg/kg/day THC). Target: for 20–40 kg, 10 mg/kg/day CBD (+0.5 THC); for >40 kg, 7.5 (+0.375). Upper limit 420 mg CBD and 21 mg THC/day. Actual mean 5.7±2.6 (extract) and 5.9±2.7 (pure) mg/kg/day +1 mg/kg/day every other day; dose split into 3 a day; tapered over 2 weeks at the end of each period	12 weeks + 4-week washout + 12 weeks	HSQ-ASD (co-primary): no difference. Much/very much improvement on CGI-I: 49% with the extract, 21% with placebo (p=0.005); 38% in the pure arm (p=0.08). SRS-2 (secondary): median improvement of 14.9 points with the extract, 3.6 with placebo (p=0.009). No difference on APSI. Sleep (CSHQ): not better than placebo (Schnapp 2022) Side effects: Sleepiness 28% extract / 23% pure / 8% placebo; reduced appetite 25% / 21% / 15%. No serious treatment-related side effects. BMI decreased with cannabinoids	PMID 33536055, doi:10.1186/s13229-021-00420-2, Sleep analysis: Schnapp 2022, PMID 35884990, doi:10.3390/biomedicines10071685, NCT02956226, placebo response analysis: Aran 2023, PMID 37176538, improvement on placebo was linked to the child understanding the purpose of the treatment, p = 0.037, and not to the family's expectations)
Silva 2024 Double-blind, placebo-controlled RCT	60 (31 active / 29 placebo) ASD, aged 5–11; refractory epilepsy excluded	CBD-rich cannabis extract (ABRACE), CBD:THC 9:1; the methods say '0.5% (5 mg/mL)', the discussion '2.5 mg/mL' (the paper's own inconsistency; the same team's 2026 sub-analysis says 0.5%)	Started at 6 drops a day (3 drops every 12 hours); not reported in mg/kg; maximum 70 drops/day; actual mean 47.4 drops (≈12 mg CBD/day according to the same team's sub-analysis) +2 drops/day twice a week; at least 1 hour apart from psychotropic medicines	12 weeks	Improvement versus placebo in a semi-structured interview: social interaction, anxiety, psychomotor agitation, number of meals. No significant difference on ATEC or CARS. In a post hoc sub-analysis of 19 children who could complete the social cognition tests (11 active / 8 placebo; ASD level 1–2), improvement in facial expression recognition and verbal theory of mind was statistically significant compared with baseline only in the active group, and the total theory of mind score improved in both groups; at the end of treatment, there was no difference between the groups on any of them (p = 0.527; 0.901; 0.967; Medeiros 2026). According to the authors, the findings do not show an effect of CBD on social cognition Side effects: According to the text, side effects in the active group 9.7% (3/31): dizziness, insomnia, colic, weight gain; in Table 2, side effects 4/31 active, 5/29 placebo	PMID 35617670, doi:10.47626/2237-6089-2021-0396, Post hoc sub-analysis: Medeiros 2026 (Autism Res, PMID 42644517, PMC13674367, 19 children who could complete the social cognition tests). Registration: ReBEC (registration no. 10743 in Silva 2024, RBR-5wr2cqq in Medeiros 2026, the same ethics approval number in both publications)
Trauner 2026 Double-blind, placebo-controlled crossover RCT (phase 2)	39 analyzed (completers: the paper says 30 in one place and 31 in another) Boys only, aged 7–14, severe behavior problems; epilepsy excluded	Pure CBD (Epidiolex); no THC, no THC detected in plasma	5 mg/kg/day (week 1) → target 20 mg/kg/day 5 → 10 → 20 mg/kg/day (weekly); 2 doses a day, with a fatty meal	8 weeks + 4-week washout + 8 weeks	No advantage over placebo on the primary outcomes (RBS-R, CBCL, ADOS-2); a marked placebo effect. Blinded clinicians correctly guessed the CBD period in 21 of the 31 children who completed (68%) (based on observed improvement; chance level in a two-period crossover design is 50%); this is not a primary outcome but a clinician impression Side effects: 118 side effects: 59 on CBD, 49 on placebo, 10 during washout. No serious treatment-related side effects. During the CBD period, 3 events of raised liver tests were recorded and led to dose changes (how many children were affected is not stated); 5 events of diarrhea and 3 of loose stools (1 of diarrhea on placebo). 4 children whose behavior worsened after CBD was stopped withdrew from the study	PMID 40410546, doi:10.1007/s10803-025-06884-y, EEG secondary analysis: Cazares 2026, PMID 41611664, NCT04517799
NCT04745026 (Jazz/GW, GWP42003-P) Double-blind, placebo-controlled parallel RCT (phase 2); no peer-reviewed publication as of October 2026, results from the ClinicalTrials.gov registration (March 30, 2025)	103 (49 CBD / 54 placebo) ASD, aged 6–17, IQ ≥70	Pure CBD oral solution, 100 mg/mL	5 mg/kg/day (1 week) → 10 mg/kg/day 5 for 1 week, then 10 mg/kg/day; 2 doses a day; tapered over 1 week at the end	12 weeks	None of the primary outcomes was significant. ABC irritability p=0.085 (numerically in favor of placebo). CGI-I response 35% versus 32% (p=0.71). Vineland-3 p=0.85. CGI-S p=0.61 Side effects: Any side effect 27/49 versus 24/54; vomiting 7 versus 4; serious side effects 2 versus 1, with 1 case of suicidal thoughts in the CBD arm; withdrawal because of side effects 2 versus 2; QTc on ECG >450 ms 7/49 versus 4/54, >500 ms 1 versus 0	NCT04745026
Parrella 2026 Double-blind, placebo-controlled crossover pilot RCT	34 randomized, 29 completed ASD, aged 5–12	CBD oil containing terpenes (Medigrowth CBD100): 100 mg/mL CBD, THC ≤1 mg/mL (as reported by the Riera 2025 review; full text could not be read)	Target 10 mg/kg/day; starting dose not reported (abstract) Not reported (abstract)	2 × 12 weeks, with an 8-week washout in between	Primary SRS-2: no difference (p=0.125). Secondary: DBC-2 social relating (p=0.024), DBC-2 anxiety (p=0.002) and parental stress APSI (p=0.044) improved. No difference on Vineland-3 Side effects: Gastrointestinal discomfort in 2 children; no other side effects reported in the abstract	PMID 41452412, doi:10.1002/aur.70159, PMID 42861636, doi:10.1002/aur.70380, registration ACTRN12622000437763
Keating 2026 (NTI164, Harmony) Double-blind, placebo-controlled RCT (phase II/III) + 8 weeks open-label; funded by the manufacturing companies (sponsor in the registration Fenix Innovation Group, partner Neurotech International); 3 of the 6 authors, including the corresponding author, work at Fenix and declared financial support from Fenix (one is also a Neurotech shareholder), 1 is a Neurotech consultant; 3 of the 4 authors who designed the trial have ties to the companies; according to the paper, the sponsors had no role in data collection or analysis; simple randomization	61 randomized, 54 completed ASD Level II/III, aged 8–17	Full-spectrum extract (NTI164): mainly CBDA, plus CBD, CBGA, CBDV and very little THC and THCA (THC <0.3% in the paper, <0.08% in the registration)	Started at 5 mg/kg/day → 20 mg/kg/day or the highest tolerated dose; actually reached 10–20 mg/kg/day +5 mg/kg/day each week; 20 reached in week 4; 2 doses a day	8 weeks	Primary CGI-S: difference −1.65 (95% CI −2.29 to −1.01; p<0.001; only the 54 people who completed the double-blind period were analyzed). Improvement in all Vineland-3 domains; improvement in some ADAMS and ASC-ASD-P subscales. SRS-2 total p=0.055. Sleep (SDSC total) p=0.214, no difference Side effects: 41 side effects in 26 participants, all mild: nausea 21.9%, headache 19.5%, abdominal pain 17.1% (percentage of side effects); similar between groups. No serious side effects; no clinically significant laboratory changes	PMID 42561512, doi:10.1016/j.neurot.2026.e01033, NCT05626959, 54 participants in the registration, no results posted)
Pretzsch 2019 (mechanism) Double-blind, placebo-controlled, single-dose, crossover	34 adult men (17 with ASD, 17 neurotypical) Adult men	Pure CBD	600 mg single dose	Single dose	Changes in glutamate and GABA on MRS; the GABA response differed in the ASD group. The authors state that it does not show effectiveness Side effects: Not reported (the full text says health checks were done, but gives no results)	PMID 30758329, doi:10.1038/s41386-019-0333-8, NCT03537950, 600 mg CBD, 600 mg CBDV and placebo, in random order): CBD with fMRI, Pretzsch 2019, PMID 31237191, especially in the ASD group), CBDV with MRS, Pretzsch 2019, PMID 31748505, Pretzsch 2021, PMID 34210360, striatal connectivity moved closer to neurotypical levels). All single-dose mechanism studies, clinical effect was not measured
Lawson 2026 Open-label phase 2, Bayesian dose-finding	23 ASD, aged 7.8–17.8, IQ ≥80	Pure CBD (Epidiolex, ≥98%)	3, 6 or 9 mg/kg/day (fixed-dose arms) Not reported (abstract)	6 weeks	Highest response 62% at 9 mg/kg/day (n=8); overall response 10/23; no response in the only child who chose sleep as the target symptom (no control group) Side effects: Related side effects: longer sleep 39%, more dreams 35%, increased saliva 30%, sedation 26%, polyuria 22%. Liver tests and blood counts were not affected	PMID 42204954, doi:10.1177/10445463261452514, NCT03900923
Hacohen 2022 Prospective, open-label	110 started, 82 completed ASD, aged 5–25 (mean 9.2)	Whole-plant extract CBD:THC 20:1 (Nitzan Spectrum, MCT oil); 5.7 mg CBD + 0.3 mg THC per drop	Started at 1 drop a day; ≤10 mg/kg/day CBD (≤400 mg/day), THC ≤0.5 mg/kg/day (≤20 mg/day) Parents increased gradually until improvement was seen (individual)	6 months	Improvement in social communication on ADOS, SRS and Vineland; no change in cognitive scores (no control group). A 6-month additional analysis by the same team (David 2025, 65 children) reported reductions in anxiety and repetitive behaviors; in a secondary analysis (Treves 2026; 81 completers out of 109 children), there was no difference in safety, effectiveness or CBD blood levels between those taking cannabis oil alone (51) and those taking it in addition to other medicines (30) Side effects: 12/110 (11%) stopped because of side effects. In an interim safety analysis of the same cohort (Stolar 2022, 59 children), no clinically significant changes in blood counts, liver enzymes, kidney, thyroid or prolactin values at month 3	PMID 36085294, doi:10.1038/s41398-022-02104-8, Additional analysis: David 2025, PMID 39047052, doi:10.1089/can.2024.0001, other publications from the same cohort (Shamir Medical Center, November 2019–May 2021), not separate studies: Stolar 2022, PMID 36249785, doi:10.3389/fphar.2022.977484, Treves 2026, PMID 40613426, doi:10.1177/02698811251332841, David 2025b, PMID 41153555, 44 of 87 families analyzed), Barchel 2026, PMID 41503912
Barchel 2018 Prospective, parent-reported; children were drawn from the licensed patient registry of the company that prepared the product (Tikun Olam), and 1 of the 8 authors works at this company; according to the paper, the company did not fund the study, and the authors declared no conflicts of interest. Some patients may overlap with Bar-Lev Schleider 2019 (both are patients of the same company)	53 ASD, aged 4–22 (median 11)	30% oil, CBD:THC 20:1	Recommended 16 mg/kg CBD (at most 600 mg/day), THC 0.8 mg/kg (at most 40 mg/day); median actually reached 90 mg/day CBD and 7 mg/day THC Not reported (not in the full text either)	Median 66 days (30–588)	Improvement according to parent reports: self-injury/meltdowns 67.6%, hyperactivity 68.4%, sleep 71.4%, anxiety 47.1%; worsening of anxiety 23.5% Side effects: Sleepiness 12/53, reduced appetite 6/53, increased appetite 4/53; all transient	PMID 30687090, doi:10.3389/fphar.2018.01521
Bar-Lev Schleider 2019 Prospective treatment program data: all autistic patients who received products in 2015–2017 from Tikun Olam, Israel's largest medical cannabis provider; the study was supported by the company, 2 of the 5 authors are company employees and 1 is a paid member of the company's scientific board. Some patients may overlap with Barchel 2018	188 ASD, mean age 12.9±7.0 years	Oil, THC:CBD 1:20 (30% CBD, 1.5% THC); 15 mg CBD + 0.75 mg THC per drop; for insomnia (46 patients), an evening oil with 3% THC was added (mean +5.0±4.5 mg/day THC)	From 1 drop 3 times a day to 20 drops 3 times a day; mean 79.5±61.5 mg CBD and 4.0±3.0 mg THC, 3 times a day; mg/kg not reported Individual; finding the right dose can take up to 2 months	6 months	Of the 93 patients assessed at month 6, marked improvement in 28 (30.1%) and moderate improvement in 50 (53.7%); survey response rate 60% Side effects: At least one side effect 25.2%: restlessness 6.6%, sleepiness 3.2%, psychoactive effect 3.2%; 5 people stopped because of side effects	PMID 30655581, doi:10.1038/s41598-018-37570-y
Fleury-Teixeira 2019 Observational (compassionate use)	18 ASD, aged 7–18; active epilepsy in 5 patients	CBD-rich extract, CBD:THC 75:1	Starting mean ~2.90 (2.30–3.60), target mean 4.55 (3.75–6.45) mg/kg/day CBD; THC mean 0.06 mg/kg/day Individual titration	6–9 months	Improvement in at least one area in 14 of 15 adherent patients; of the 5 patients with epilepsy, seizures fell by 50% in 3 and disappeared in 2 (no control) Side effects: 3/18 (17%) stopped within the first month because of side effects; all were taking multiple medicines, including antipsychotics. In the others, sleepiness and irritability (3 cases each)	PMID 31736860, doi:10.3389/fneur.2019.01145
Fortini 2025 / 2026 Prospective before-and-after, add-on treatment	20 (13 patients completed long-term follow-up) Severe ASD + intellectual disability, aged 3–18	Pure CBD	Median total 363.5 mg/day (100–700); mg/kg not reported Not reported (abstract)	Median 11 months; long-term follow-up mean 27.6±1.3 months	Improvement in at least one parent-selected symptom in 18/20 patients; about 30% improvement in irritability, social withdrawal and hyperactivity; concomitant medicines could be reduced in 40% (no control) Side effects: Side effects in 65% (13/20): increased irritability, reduced appetite; mild or moderate, transient	PMID 39965749, doi:10.1016/j.pbb.2025.173971, PMID 42248366, doi:10.1016/j.pbb.2026.174220
Erridge 2022; Aggarwal 2026 (UK Medical Cannabis Registry) Uncontrolled case series (registry)	74; 130 Adults with ASD (mean age 32.7)	Mostly THC-dominant dried flower and oils	CBD median 8.50 mg/day, THC median 20 mg/day (2026); at month 18, THC median 192.25 mg/day, CBD 26.25 mg/day; in the 2022 cohort, THC median 112.5 mg/day Titration by a specialist physician	6 months; 18 months	Linked to improvement in GAD-7, sleep and EQ-5D-5L; no control group. In the 2026 cohort, the effect wanes after month 3; the authors discuss THC tolerance. Note on interests: according to the journal's correction, most authors of the 2026 publication work at a private medical cannabis clinic Side effects: 180 side effects in 14/74 people (2022). 232 side effects in 25/130 people; 42 severe, 3 life-threatening or disabling (2026)	PMID 36159065, doi:10.1177/20451253221116240, PMID 42387975, doi:10.1002/npr2.70146
Aran 2019 Retrospective	60 ASD, aged 5–17.5	CBD-rich cannabis, CBD:THC 20:1 (Bilge 2021 review; Aran 2021 writes that it took the same ratio from earlier open studies)	According to the Bilge 2021 review: in those taking 3 doses a day (n=44), mean 3.8±2.6 CBD and 0.29±0.22 THC mg/kg/day; in those taking 2 doses a day (n=16), 1.8±1.6 CBD and 0.22±0.14 THC mg/kg/day According to the review, 2–4 weeks	According to the review, mean follow-up 10.9±2.3 months	Much/very much improvement in behavioral outbreaks 61% (caregiver global impression) Side effects: Sleep disturbance 14%, irritability 9%, loss of appetite 9%. In a 13-year-old girl taking more THC (6.5 mg/kg/day CBD; when THC rose to 0.72 mg/kg/day), a transient, serious psychotic event; an antipsychotic was needed (as reported by Bilge 2021)	PMID 30382443, doi:10.1007/s10803-018-3808-2
Bilge & Ekici 2021 (Turkey) Retrospective, single center	33 ASD, mean age 7.7±5.5 years; all were taking antipsychotics	Full-spectrum CBD, trace THC (CBDistillery, CBDodgamax; 500/1000/2500 mg/30 mL drops); ratio not reported. The abstract gives THC as '<3%', the methods as the maximum allowed in Turkey, 0.3%	Started at 1 drop twice a day (according to body weight); mean 0.7 (0.3–2) mg/kg/day; daily maintenance CBD did not exceed 40 mg in any patient +1 drop every 3 days	Median 6.5 months (3–28)	Parent reports: 32.2% reduction in behavior problems, 22.5% increase in expressive language; the antipsychotic could be stopped in only 1 patient Side effects: Restlessness 7/31 (22%), which subsided when the dose was lowered; a generalized seizure in a 13-year-old boy after 5 mg CBD (the drug was stopped); stopped in week 2 in one child because of increased stereotypy; no laboratory changes	PMID 34911567, doi:10.1186/s42238-021-00108-7
Wilson 2026 Systematic review and meta-analysis (54 RCTs in mental disorders)	54 RCTs; 2,477 participants (all mental disorders); 2 RCTs in the ASD analysis (from the journal correction) Mental disorders (ASD sub-analysis)	Various cannabinoids; both ASD studies used THC and CBD together			In ASD, reduction in autistic traits SMD −0.36 (−0.66 to −0.07; 2 RCTs, both with products containing THC and CBD together); quality of evidence generally low. The correction published in the journal (June 16, 2026) corrected the number and type of the ASD studies; the result did not change Side effects: Risk of any side effect increased: OR 1.75 (NNTH=7); serious side effects did not increase	PMID 41856154, doi:10.1016/S2215-0366(26)00015-5, Correction: Lancet Psychiatry 2026, 13(8):e11, PMID 42302809, 2026, the axis of a figure and the study counts in the table were corrected, results unchanged), PROSPERO CRD42023392718
Vasconcelos 2026 Systematic review (GRADE); 4 RCTs + 8 non-RCTs	not reported ASD in children and adolescents	Various cannabinoid products	Titration generally up to ~10 mg/kg/day		In a single RCT (20:1 extract), improvement in CGI and social communication; the other standard measures were negative; no difference in sleep or overall severity; non-RCT studies at very low certainty	PMID 41974303, doi:10.1016/j.pnpbp.2026.111697
Riera 2025 Systematic review (Cochrane methods); 11 studies: 4 completed RCTs, 7 ongoing	351 (4 completed RCTs) ASD	Various cannabis derivatives			Certainty very low to low; no study measured quality of life	PMID 40605143, doi:10.1002/jcph.70068
Köck 2024 Systematic review and meta-analysis of RCTs	8 RCTs, 474 participants (ASD: 2 RCTs) Psychiatric disorders in young people	Various cannabinoids			ASD: g=0.264 (0.107–0.421; 2 RCTs: Aran 2021, Silva 2024); across all disorders, CBD dose was not related to effect size (ρ=−0.014)	PMID 39696457, doi:10.1186/s13034-024-00846-5
Berry-Kravis 2022 (CONNECT-FX) Double-blind, placebo-controlled RCT (phase 3; 21 centers: US, Australia, New Zealand); 2-week single-blind placebo run-in; fragile X syndrome. Funded by the manufacturer (Zynerba); all 14 authors declared ties to the manufacturer (research funding, consulting; 3 employees, 1 former employee)	212 randomized (110 CBD, 102 placebo); 211 in the safety analysis, 210 in the efficacy analysis Fragile X syndrome (FMR1 full mutation), aged 3–17 (mean 9.7), 75% male; high social avoidance or irritability scores; at most 2 add-on psychiatric medicines	ZYN002: laboratory-made (not plant-derived) CBD gel applied to the skin; contains no THC	250 mg CBD a day for those weighing ≤35 kg, 500 mg for those >35 kg (split into two doses, on the skin) None; fixed dose by weight	12 weeks	Primary outcome (ABC-CFXS social avoidance, week 12) showed no difference: mean change −2.68 versus −2.29 on placebo (p = 0.321). Key secondary outcomes also showed no difference: irritability p = 0.149, social unresponsiveness p = 0.607, any improvement on CGI-I 46.8% versus 40.5% (p = 0.376). In a subgroup analysis that the paper calls 'ad hoc' and that was not in the original analysis plan (according to the registration, defined before unblinding), the difference in social avoidance in 169 children with at least 90% methylation of the FMR1 gene was −1.00 (nominal p = 0.020; no correction for multiple comparisons); in the small group with methylation below 90%, the effect went in the opposite direction. This finding is exploratory and does not count as proof of effect. The second RCT testing this subgroup as its main measure (RECONNECT, NCT04977986, 257 people) ended in August 2025; as of October 2026, its results are not in the registry or in a peer-reviewed journal Side effects: Any side effect 57.8% versus 50.0% on placebo; all mild or moderate; no serious side effects; the most common treatment-related side effect was application-site pain (6.4% versus 1.0%)	PMID 36434514, doi:10.1186/s11689-022-09466-6, NCT03614663, confirmatory study RECONNECT, NCT04977986, ended August 29, 2025, as of October 2026, no results in the registry and no peer-reviewed publication)
Heussler 2019 (FAB-C) Open-label phase 1/2, multicenter (Australia); no control group; fragile X syndrome. Funded by the manufacturer (Zynerba), which took part in the design and interpretation; 5 of the 8 authors are linked to the manufacturer (3 employees, 1 former employee, 1 consultant)	20 started, 18 completed Fragile X syndrome, aged 6–17 (mean 10.4), 75% male; high anxiety scores	ZYN002, synthetic CBD gel applied to the skin	Started at 50 mg a day, at most 250 mg; of the 18 who completed, 16 received 250 mg and 2 received 100 mg 50 → 100 → 250 mg/day over 6 weeks (at the investigator's discretion)	12 weeks (6 weeks of dose adjustment + 6 weeks at a fixed dose); 13 people moved to the 24-month extension	The primary ADAMS total score (anxiety, mood, behavior) fell from an average of 32.1 to 18.1 (p < 0.001); improvement also in anxiety (PARS-R), ABC-CFXS subscales and quality of life; no significant change in depressed mood. No control group Side effects: At least one side effect in 17 of 20 people (85%), 70% of them mild; no serious side effects; side effects possibly related to treatment in 6 people (dryness or rash at the application site, hyperactivity, stereotypy, nightmares); most common were gastroenteritis 25% and vomiting 10%; 1 person stopped because of worsening eczema	PMID 31370779, doi:10.1186/s11689-019-9277-x, ACTRN12617000150347
Berry-Kravis 2025 (fragile X extension) Interim analysis of an open-label extension study (ZYN2-CL-017; data cutoff January 31, 2024); primary aim safety; no control group. Funded by the manufacturer (Harmony Biosciences); all 15 authors declared ties to the manufacturer (6 employees)	240 (10 from FAB-C, 197 who completed CONNECT-FX, 33 who could not be enrolled in CONNECT-FX); 196 in the efficacy analysis Fragile X syndrome, aged 3–17 at entry (mean 9.7), 76% male	ZYN002, synthetic CBD gel applied to the skin	250 mg a day (≤35 kg) or 500 mg (>35 kg); after month 1, between 125 and 750 mg at the investigator's discretion Individual (increase or decrease with the approval of the investigator and medical monitor)	Average 28 months including the earlier studies; at the data cutoff, 75 people were continuing and 165 had left	Safety (primary): treatment-related side effects 12.9%; serious side effects 4.6% (none considered drug-related); stopping because of side effects 3.3%. Secondary: compared with the CONNECT-FX baseline, the social avoidance score fell by −3.37 at month 1 in the group continuing the gel and by −3.36 at month 3 in the group switching from placebo, and was maintained in the following months. No control group; the dropout rate is high (165 of 240 people), and according to the authors, good responders staying in the study may make the result look optimistic Side effects: At least one side effect in 68.3% of 240 people (98% mild to moderate); most common were upper respiratory infection 20%, COVID-19 12.5%, vomiting 7.9%, fever 7.1%, application-site pain 6.7%; ALT above 3 times normal in 3 people (1.3%); events leading to stopping include aggression, behavior disorder, irritability, suicidal thoughts and fatty liver	PMID 41254489, doi:10.1186/s11689-025-09657-x, NCT03802799, no results as of October 2026)
Treves 2025 (Israeli national data) Retrospective national cohort (Israeli Medical Cannabis Agency data, 2018–2022) and data from the Shamir Medical Center open-label study (the same group of children as Hacohen 2022; 59 children in the methods, 110 in the flow chart); aim: staying on treatment. The authors declared no conflicts of interest	1,341 children under 12; 751 with ASD (367 children with ASD in the one-year follow-up analysis) Children under 12 receiving medical cannabis oil (ASD, epilepsy, Tourette syndrome, cancer); adolescents and young adults for comparison	Licensed medical cannabis oils; most children with ASD received a CBD-rich product that also contained THC (THC:CBD mostly 1:10 or lower; in older children, the THC ratio rose over time)	No mg/kg in the national data (amount dispensed per month) Each month during treatment, 15–25% of patients changed the amount and 10–20% changed the THC:CBD ratio	First 3 months; 12 months in ASD	In children with ASD, the share who continued treatment for one year was about 26% both in the national data and in the open-label study; in the national data, 48% completed 6 months (about 75% in the open-label study). In the open-label study, the main reasons for stopping were side effects and lack of effect, and in the long term also bureaucratic and logistical difficulties. Effectiveness was not measured; no control group Side effects: In the open-label study, the main reasons for dose reductions were psychiatric, neurological and general side effects (tiredness, restlessness, sleep problems, aggression; the reports also include suicidal thoughts); in one third of reductions, no reason was recorded	PMID 41357860, doi:10.3389/fphar.2025.1646560

Reported side effects

Sleepiness: 28% (20:1 extract) / 23% (pure) / 8% (placebo) (Aran 2021)
Reduced appetite: 25% / 21% / 15% (Aran 2021); BMI decreased in the cannabinoid arms
Sleep disturbance/insomnia, irritability, restlessness (Aran 2019; Bar-Lev Schleider 2019; Fortini 2025)
Vomiting (NCT04745026: 7 versus 4); gastrointestinal complaints (Parrella 2026; Keating 2026)
With pure CBD, longer sleep 39%, sedation 26%, increased saliva 30%, polyuria 22% (Lawson 2026)
Worsening behavior: in Fleury-Teixeira 2019, 3/18 stopped early; in Hacohen 2022, 12/110 stopped because of side effects; in Trauner 2026, 4 children got worse after CBD was stopped
Serious THC-related psychiatric events: a psychotic event in a girl in Aran 2019; mania or psychosis in three young adult men using CBD+THC in Al-Soleiti 2021 (PMID 34505186)
1 serious side effect in the pure CBD arm: suicidal thoughts (NCT04745026)
Raised liver enzymes with 20 mg/kg/day pure CBD led to dose changes (Trauner 2026; 3 events during the CBD period)
QTc prolongation on ECG: >450 ms 7/49 versus 4/54, >500 ms 1 versus 0 (NCT04745026)
With pure CBD gel applied to the skin (fragile X): application-site pain 6.4% versus 1.0% on placebo, no serious side effects (Berry-Kravis 2022); in an open extension averaging 28 months, treatment-related side effects 12.9%, ALT above 3 times normal in 3 of 240 people (Berry-Kravis 2025)

Monitoring done in studies

At the start, liver and kidney function, complete blood count and fasting blood sugar (Silva 2024)
Every two weeks, complete blood count, comprehensive metabolic panel, cannabinoid levels and ECG (Trauner 2026)
At the start and at month 3, complete blood count, liver enzymes, kidney, lipids, thyroid and prolactin (interim safety analysis of the Hacohen 2022 cohort: Stolar 2022, PMID 36249785)
Safety blood tests at the start and at weeks 4, 8 and 16 (Keating 2026)
Asking about side effects with a structured scale or by phone every 2 weeks (Aran 2021; Barchel 2018; Hacohen 2022)
With CBD gel applied to the skin, a skin examination of the application site at every visit; also ECG, safety blood tests and screening for suicidal thoughts (Berry-Kravis 2022)

Who should be careful

History of psychosis or psychosis in the family: excluded from studies (Hacohen 2022); Barchel 2018 considers a history of psychotic episodes a contraindication; cases of psychosis and mania have been reported with THC-containing products
Bipolar disorder, active major depression, history of suicidal behavior (NCT04745026 exclusion criteria)
Liver dysfunction: NCT04745026 excluded those with ALT/AST >2×ULN, and Trauner 2026 excluded those with liver disease
People taking valproate or clobazam were excluded because of interactions (Trauner 2026)
Pregnancy and breastfeeding (Hacohen 2022 and NCT04745026 exclusion criteria)
Allergy to product ingredients, e.g. sesame oil (Trauner 2026; NCT04745026)
History of substance misuse (NCT04745026; Hacohen 2022)
Children taking antipsychotics or multiple medicines: close monitoring; current medicines should not be stopped suddenly (Fleury-Teixeira 2019)
Israeli context: a prescription requires two FDA-approved antipsychotics to have been tried first; in Clalit data, this was documented in only 28% of prescriptions, and 4.3% were written for children under 5 (Sadeh 2026, PMID 41334706)

Drug interactions

Antiepileptics (epilepsy cohort, not specific to ASD): as the CBD dose went up, N-desmethylclobazam, topiramate and rufinamide levels rose and the clobazam level fell; in adults, zonisamide and eslicarbazepine rose; with valproate, AST/ALT were high (Gaston 2017, PMID 28782097)
CYP: CBD is broken down by CYP2C19 and CYP3A4, THC by CYP2C9 and CYP3A4; ketoconazole raised CBD and THC levels with the THC/CBD mouth spray; rifampin lowered exposure to approved oral CBD by ~32% (FDA label), and with low-dose THC/CBD mouth spray the drop was ~50–60% (Stott 2013, PMID 23750331)
Barchel 2018 discussion (not its own data; citing a review in the context of products that also contain THC): phenytoin levels may rise; possible interaction with SSRIs, tricyclic antidepressants and CNS depressants. For pure CBD, the EMA product information states that no dose adjustment is needed for CYP2C9 substrates such as phenytoin
CBD blood levels were lower on average in those taking antipsychotics, SSRIs, alpha-2 agonists, stimulants or mood stabilizers (Trauner 2026)
All 3 patients who had serious side effects were taking multiple medicines, including antipsychotics; in two, the antipsychotic had been stopped suddenly without supervision (Fleury-Teixeira 2019)
Barchel 2018, based on animal studies, reports that THC may reverse the effect of risperidone
Sleepiness was more common in those taking chronic medication (Aran 2021)
CNS depressants and anticholinergics: effects such as drowsiness and ataxia may add up (Campos 2024, PMID 38794183)
NCT04745026 excluded people taking these medicines (not evidence of interaction; a sponsor safety limit): clobazam; omeprazole, lansoprazole; tolbutamide; warfarin; sirolimus, everolimus, temsirolimus, tacrolimus; repaglinide, pioglitazone, rosiglitazone; montelukast, bupropion, efavirenz

Abbreviations: RCT, randomized controlled trial (a study in which participants are assigned to groups by chance); placebo, an inactive substance; SMD, OR and RR, effect measures used in analyses that pool studies; CYP, liver enzymes that break down medicines.

If you are considering CBD

You can talk openly with your child's doctor about your expectations and which symptom you want to track.
Review all the medicines your child takes together, especially antipsychotics, antidepressants and epilepsy medicines, for interactions; do not stop current medicines suddenly.
If there is a history of psychosis in your child or your family, be sure to mention it; cases of psychosis and mania have been reported with THC-containing products.
If a product is being considered, go over the CBD and THC amounts on its label and its certificate of analysis (COA) with the doctor; plan monitoring with liver tests and blood counts.

Choose your age group, medicines and situation to prepare a short summary you can take to your doctor. Nothing you enter is sent anywhere.

Open the personal warnings tool Convert drops ↔ mg

Bottom line

CBD in autism is a topic being studied with great interest, but the answer is not yet clear. Although some families describe improvement, in controlled studies the effect of pure CBD did not clearly stand apart from placebo. Making decisions together with a doctor who knows your child is the safest path.

Frequently asked questions

Does CBD oil work for autism?

The evidence on this is weak. In the two controlled studies with pure CBD, there was no clear difference from a dummy medicine (placebo) in behavior problems. Some studies with extracts that also contain THC saw improvement (the most positive result came from a study funded by the company that makes the product), but the results were not consistent, and THC has its own risks. In fragile X syndrome, the most common single-gene cause of autism, pure CBD gel applied to the skin was also not clearly different from placebo on the main measure in a large controlled study.

Can CBD oil be used in children?

Most autism studies were done in children and teens, under a doctor's supervision and with regular blood tests. There is no approved CBD medicine for autism. Because of risks such as sleepiness, reduced appetite and interactions with other medicines, the decision should be made together with your child's doctor.

Does CBD help with sleep problems in autism?

In two controlled studies, CBD products were not found to be better than placebo for sleep problems. Some families' accounts reported improved sleep, but those studies had no comparison group.

Why don't the improvements families see count as enough evidence?

The placebo effect is strong in autism studies: children taking a dummy medicine can also improve clearly. That is why observations without a comparison group do not count as enough evidence on their own.

Can CBD oil be obtained with a prescription for autism in Turkey?

In Turkey, products derived from cannabis are regulated by the Kenevirden Elde Edilen Ürünlere Dair Yönetmelik (Regulation on Products Derived from Hemp) dated January 31, 2026. These products, including CBD, can be sold only in pharmacies. Those classified as medicines and health products are dispensed with a prescription; the regulation does not state a prescription requirement for supplement products and cosmetics with a THC content of at most 0.3%. As of October 2026, there is no approved CBD medicine in Turkey. If you are thinking of using it, talk to your doctor first.

Does CBD work for fragile X syndrome?

The evidence on this is weak. In fragile X syndrome, pure CBD gel applied to the skin was not clearly different from placebo on the main measure, social avoidance, in a controlled study of 212 children. In an analysis added afterward, a small difference was seen in children in whom the gene is most likely silenced; however, this is an exploratory finding, and the results of the second large study in these children had not been published as of October 2026. In the studies, the gel was generally well tolerated; the most common treatment-related side effect was pain or redness where it was applied. This gel is an investigational medicine; you can talk to your doctor about any product you are considering for your child.

Has CBD been tried for behavior problems in my child with intellectual disability?

There is not enough research on this question yet. In children with intellectual disability and severe behavior problems, pure CBD has been compared with placebo only in an 8-child pilot study; this study was designed not to measure the effect but to test feasibility, and all the children taking CBD were autistic. There is no controlled study giving a separate result for intellectual disability without autism. The results of the large controlled study with 132 children had not been published as of October 2026. Because the medicines your child takes (especially antipsychotics and epilepsy medicines) may interact with CBD, you can talk about this with your child's doctor.

References

Show 24 references

Aran 2021. Double-blind, placebo-controlled RCT, 3 arms, prespecified crossover (efficacy from period 1 only). PMID 33536055, doi:10.1186/s13229-021-00420-2, Sleep analysis: Schnapp 2022, PMID 35884990, doi:10.3390/biomedicines10071685, NCT02956226, placebo response analysis: Aran 2023, PMID 37176538, improvement on placebo was linked to the child understanding the purpose of the treatment, p = 0.037, and not to the family's expectations)
Silva 2024. Double-blind, placebo-controlled RCT. PMID 35617670, doi:10.47626/2237-6089-2021-0396, Post hoc sub-analysis: Medeiros 2026 (Autism Res, PMID 42644517, PMC13674367, 19 children who could complete the social cognition tests). Registration: ReBEC (registration no. 10743 in Silva 2024, RBR-5wr2cqq in Medeiros 2026, the same ethics approval number in both publications)
Trauner 2026. Double-blind, placebo-controlled crossover RCT (phase 2). PMID 40410546, doi:10.1007/s10803-025-06884-y, EEG secondary analysis: Cazares 2026, PMID 41611664, NCT04517799
NCT04745026 (Jazz/GW, GWP42003-P). Double-blind, placebo-controlled parallel RCT (phase 2); no peer-reviewed publication as of October 2026, results from the ClinicalTrials.gov registration (March 30, 2025). NCT04745026
Parrella 2026. Double-blind, placebo-controlled crossover pilot RCT. PMID 41452412, doi:10.1002/aur.70159, PMID 42861636, doi:10.1002/aur.70380, registration ACTRN12622000437763
Keating 2026 (NTI164, Harmony). Double-blind, placebo-controlled RCT (phase II/III) + 8 weeks open-label; funded by the manufacturing companies (sponsor in the registration Fenix Innovation Group, partner Neurotech International); 3 of the 6 authors, including the corresponding author, work at Fenix and declared financial support from Fenix (one is also a Neurotech shareholder), 1 is a Neurotech consultant; 3 of the 4 authors who designed the trial have ties to the companies; according to the paper, the sponsors had no role in data collection or analysis; simple randomization. PMID 42561512, doi:10.1016/j.neurot.2026.e01033, NCT05626959, 54 participants in the registration, no results posted)
Pretzsch 2019 (mechanism). Double-blind, placebo-controlled, single-dose, crossover. PMID 30758329, doi:10.1038/s41386-019-0333-8, NCT03537950, 600 mg CBD, 600 mg CBDV and placebo, in random order): CBD with fMRI, Pretzsch 2019, PMID 31237191, especially in the ASD group), CBDV with MRS, Pretzsch 2019, PMID 31748505, Pretzsch 2021, PMID 34210360, striatal connectivity moved closer to neurotypical levels). All single-dose mechanism studies, clinical effect was not measured
Lawson 2026. Open-label phase 2, Bayesian dose-finding. PMID 42204954, doi:10.1177/10445463261452514, NCT03900923
Hacohen 2022. Prospective, open-label. PMID 36085294, doi:10.1038/s41398-022-02104-8, Additional analysis: David 2025, PMID 39047052, doi:10.1089/can.2024.0001, other publications from the same cohort (Shamir Medical Center, November 2019–May 2021), not separate studies: Stolar 2022, PMID 36249785, doi:10.3389/fphar.2022.977484, Treves 2026, PMID 40613426, doi:10.1177/02698811251332841, David 2025b, PMID 41153555, 44 of 87 families analyzed), Barchel 2026, PMID 41503912
Barchel 2018. Prospective, parent-reported; children were drawn from the licensed patient registry of the company that prepared the product (Tikun Olam), and 1 of the 8 authors works at this company; according to the paper, the company did not fund the study, and the authors declared no conflicts of interest. Some patients may overlap with Bar-Lev Schleider 2019 (both are patients of the same company). PMID 30687090, doi:10.3389/fphar.2018.01521
Bar-Lev Schleider 2019. Prospective treatment program data: all autistic patients who received products in 2015–2017 from Tikun Olam, Israel's largest medical cannabis provider; the study was supported by the company, 2 of the 5 authors are company employees and 1 is a paid member of the company's scientific board. Some patients may overlap with Barchel 2018. PMID 30655581, doi:10.1038/s41598-018-37570-y
Fleury-Teixeira 2019. Observational (compassionate use). PMID 31736860, doi:10.3389/fneur.2019.01145
Fortini 2025 / 2026. Prospective before-and-after, add-on treatment. PMID 39965749, doi:10.1016/j.pbb.2025.173971, PMID 42248366, doi:10.1016/j.pbb.2026.174220
Erridge 2022; Aggarwal 2026 (UK Medical Cannabis Registry). Uncontrolled case series (registry). PMID 36159065, doi:10.1177/20451253221116240, PMID 42387975, doi:10.1002/npr2.70146
Aran 2019. Retrospective. PMID 30382443, doi:10.1007/s10803-018-3808-2
Bilge & Ekici 2021 (Turkey). Retrospective, single center. PMID 34911567, doi:10.1186/s42238-021-00108-7
Wilson 2026. Systematic review and meta-analysis (54 RCTs in mental disorders). PMID 41856154, doi:10.1016/S2215-0366(26)00015-5, Correction: Lancet Psychiatry 2026, 13(8):e11, PMID 42302809, 2026, the axis of a figure and the study counts in the table were corrected, results unchanged), PROSPERO CRD42023392718
Vasconcelos 2026. Systematic review (GRADE); 4 RCTs + 8 non-RCTs. PMID 41974303, doi:10.1016/j.pnpbp.2026.111697
Riera 2025. Systematic review (Cochrane methods); 11 studies: 4 completed RCTs, 7 ongoing. PMID 40605143, doi:10.1002/jcph.70068
Köck 2024. Systematic review and meta-analysis of RCTs. PMID 39696457, doi:10.1186/s13034-024-00846-5
Berry-Kravis 2022 (CONNECT-FX). Double-blind, placebo-controlled RCT (phase 3; 21 centers: US, Australia, New Zealand); 2-week single-blind placebo run-in; fragile X syndrome. Funded by the manufacturer (Zynerba); all 14 authors declared ties to the manufacturer (research funding, consulting; 3 employees, 1 former employee). PMID 36434514, doi:10.1186/s11689-022-09466-6, NCT03614663, confirmatory study RECONNECT, NCT04977986, ended August 29, 2025, as of October 2026, no results in the registry and no peer-reviewed publication)
Heussler 2019 (FAB-C). Open-label phase 1/2, multicenter (Australia); no control group; fragile X syndrome. Funded by the manufacturer (Zynerba), which took part in the design and interpretation; 5 of the 8 authors are linked to the manufacturer (3 employees, 1 former employee, 1 consultant). PMID 31370779, doi:10.1186/s11689-019-9277-x, ACTRN12617000150347
Berry-Kravis 2025 (fragile X extension). Interim analysis of an open-label extension study (ZYN2-CL-017; data cutoff January 31, 2024); primary aim safety; no control group. Funded by the manufacturer (Harmony Biosciences); all 15 authors declared ties to the manufacturer (6 employees). PMID 41254489, doi:10.1186/s11689-025-09657-x, NCT03802799, no results as of October 2026)
Treves 2025 (Israeli national data). Retrospective national cohort (Israeli Medical Cannabis Agency data, 2018–2022) and data from the Shamir Medical Center open-label study (the same group of children as Hacohen 2022; 59 children in the methods, 110 in the flow chart); aim: staying on treatment. The authors declared no conflicts of interest. PMID 41357860, doi:10.3389/fphar.2025.1646560

Points that could not be verified from the full text

The full text of Aran 2019 is not open access; dose and duration are taken from the Bilge 2021 review (Aran 2021 also confirms the 20:1 ratio)
CASCADE (NCT04520685; 86 people, pure CBD 10 mg/kg/day, primary data collection ended in January 2024): status in the registry is 'unknown' (last updated May 2024), no results posted, only a methods paper; risk of publication bias
The full text of Parrella 2026 could not be read (publisher access blocked); the starting dose and titration are not in the abstract or the registration; THC content only as reported by the Riera 2025 review (≤1 mg/mL)
That David 2025 (PMID 39047052) comes from the same group of children as Hacohen 2022 was inferred from the author team and the description in the abstract; the full text could not be read
RECONNECT (NCT04977986; pure CBD gel applied to the skin in fragile X, 257 people, ended in August 2025): as of October 2026, the results had not been published either in the study registry or in a peer-reviewed journal. A page of FRAXA, the fragile X patient association, reports that the company running the study announced on September 24, 2025 that there was no difference from placebo on the main measure; this is a secondhand report, contains no numbers and could not be verified (both the company's own announcement and the association's page gave an access block on a second reading). For this reason, it is not stated as a result on the page.
The Australian study comparing pure CBD with placebo in 132 children with intellectual disability and severe behavior problems (NCT04821856; 10 mg of pure CBD per kilogram of body weight a day, 8 weeks; children with autism are not excluded) ended in August 2025; as of October 2026, there are no results in the registry and no publication. The same team's 8-child pilot study (Efron 2021) is now in the table; its full text could not be read: the efficacy figures are not in the abstract, and the children's diagnoses (all children in the CBD group were autistic, 1 in 4 in the placebo group) and side effects were taken from a review in which the team took part (Rice 2024).
A 20-person pilot study without a comparison group with nabilone, a synthetic THC-like drug, for severe behavior problems (aggression) in adults with intellectual disability (NCT05273320, Canada; protocol: Lin 2023, PMID 37043511) was completed in August 2025; as of October 2026, there are no results or publication in the registry. In the Netherlands, single-patient crossover trials with pure CBD in 30 patients for severe behavior problems in tuberous sclerosis, MPS III and fragile X syndrome began in January 2025 (protocol: Müller 2024, PMID 38177999; EudraCT 2021-003250-23); in the EU clinical trials register in October 2026 the status was 'ongoing', with no results; no results publication was found on PubMed.

Research history

Here we record what changed on this page, when and why, with dates. The newest change is at the top.

October 31, 2026
A separate section was added to the page for children with intellectual disability and severe behavior problems. The only controlled study testing pure CBD on this question is an 8-child pilot study (Efron 2021): its aim was to test whether such a study is feasible; the authors reported a signal in favor of CBD, but the number of children was far too small for a conclusion, and all the children taking CBD were autistic. There is no controlled study giving a separate result for intellectual disability without autism. The results of the large 132-child study and of a 20-person pilot study with nabilone in adults had not been published as of October 2026. In autism, the result and the strength of evidence did not change.
Related studies: (PMID 32478863, doi:10.1111/bcp.14399)
October 30, 2026
The last sentence of the page's short summary was rewritten in calmer, more guiding language: instead of "Talk to your doctor before you start.", it now says who you can talk with, and about what, if you are considering CBD. The result and the strength of evidence did not change.
October 30, 2026
The correction to the Wilson 2026 meta-analysis published in the journal was read: the results did not change. According to the corrected table, the meta-analysis's autism result rests on 2 studies done with products containing both THC and CBD; there is no study done with CBD alone. This information was added to the page.
Related studies: Wilson 2026, published April 2026 (PMID 41856154, doi:10.1016/S2215-0366(26)00015-5, Correction: Lancet Psychiatry 2026, 13(8):e11, PMID 42302809, 2026, the axis of a figure and the study counts in the table were corrected, results unchanged), PROSPERO CRD42023392718)

Older changes (8)

October 28, 2026
October 2026 additional review: sources found for this page during searches for other topics were read and added to the page. A later sub-analysis (Medeiros 2026) of the 60-child controlled study in Brazil (Silva 2024) had been listed as a separate row in the table; because it is a second publication of the same trial, it is no longer a separate row, and its finding (no difference from placebo in facial expression recognition and theory of mind in 19 children who completed the social cognition tests) was moved to the main study's row. The number of controlled studies did not change; the conclusion and the strength of evidence did not change.
October 28, 2026
Evidence changed: The finding for CBD in autism was changed from "Mixed results" to "No effect found"; the strength of evidence (Weak) and the strength of evidence for THC-containing products (Weak) did not change. On this site, the CBD label is given according to the evidence for pure CBD. In the two controlled studies with pure CBD in autism (39 and 103 children; the results of the second only in the ClinicalTrials.gov registration), the main measures of behavior problems were no different from placebo. The positive controlled findings that gave the "mixed" impression came from extracts that also contain THC or have an unclear THC amount, and mostly from secondary measures; these are now described under the evidence for THC-containing products. Positive results for pure CBD were seen only in small studies without a control group. The fragile X syndrome studies are described in a separate section and were not used as a basis for this label. Because the studies are few, small and short, the evidence is weak.
Studies that led to the change: Trauner 2026, published November 2026 (online May 24, 2025) (PMID 40410546, doi:10.1007/s10803-025-06884-y, EEG secondary analysis: Cazares 2026, PMID 41611664, NCT04517799); NCT04745026 (Jazz/GW, GWP42003-P) (NCT04745026)
October 28, 2026
In October 2026, PubMed (835 records), Cochrane and ClinicalTrials.gov (87 records) were systematically searched again. The scope was widened and a fragile X syndrome section was added to the page: a 212-child controlled study with pure CBD gel applied to the skin (no different from placebo on the primary outcome; a difference only in an exploratory subgroup analysis added afterward), a 20-child open-label study and a 240-child long-term open extension study. A second large controlled study in this subgroup has ended, but its results have not been published; because the company's announcement about this study could not be verified, it was mentioned only under unverifiable points. A real-world study from Israel's national data was also added (share of children who continued treatment for a year about 26%). The result for THC-containing products was rewritten to describe all four controlled studies. Corrected: the upper limit of the average doses reached in the studies (5.9, not 7.88 mg/kg/day) and that in the controlled studies with pure CBD the dose was fixed at 10–20 mg/kg/day; two publications from the same Israeli group of children were not counted as separate studies and were moved to the row of the main study (Hacohen 2022); the other publications of the single-dose brain imaging study were added; a study's company ties, confidence interval and number of children analyzed; a meta-analysis's number of participants and correction notice; a publication's year. The strength of evidence did not change.
Related studies: Berry-Kravis 2022 (CONNECT-FX), published November 25, 2022 (PMID 36434514, doi:10.1186/s11689-022-09466-6, NCT03614663, confirmatory study RECONNECT, NCT04977986, ended August 29, 2025, as of October 2026, no results in the registry and no peer-reviewed publication)); Heussler 2019 (FAB-C), published August 2, 2019 (PMID 31370779, doi:10.1186/s11689-019-9277-x, ACTRN12617000150347); Berry-Kravis 2025 (fragile X extension), published November 18, 2025 (PMID 41254489, doi:10.1186/s11689-025-09657-x, NCT03802799, no results as of October 2026)); Treves 2025 (Israeli national data), published November 21, 2025 (PMID 41357860, doi:10.3389/fphar.2025.1646560)
October 27, 2026
Where the 'Mixed results' label comes from was written into the plain summary and the evidence summary: in the controlled studies of pure CBD, the main outcomes were negative, and the positive findings came with products that also contain THC and mostly on secondary measures. The conclusion and the strength of evidence did not change.
October 27, 2026
The year of a controlled study with pure CBD was corrected to the journal's print year (Trauner 2026), and a note was added that the clinician impression in this study was not a primary outcome; the authors' declaration of interests was added to the UK registry study. The conclusion and the strength of evidence did not change.
October 21, 2026
Newly published and previously missed clinical studies were searched in PubMed and ClinicalTrials.gov. A new publication from the Hacohen 2022 study team (David 2025) was not counted as a separate study; it was added as a note to that study's row. The conclusion and the strength of evidence did not change.
Related studies: Hacohen 2022, published September 9, 2022 (PMID 36085294, doi:10.1038/s41398-022-02104-8, Additional analysis: David 2025, PMID 39047052, doi:10.1089/can.2024.0001, other publications from the same cohort (Shamir Medical Center, November 2019–May 2021), not separate studies: Stolar 2022, PMID 36249785, doi:10.3389/fphar.2022.977484, Treves 2026, PMID 40613426, doi:10.1177/02698811251332841, David 2025b, PMID 41153555, 44 of 87 families analyzed), Barchel 2026, PMID 41503912)
October 21, 2026
The full texts of the studies were re-read and the information on the page was checked against its sources. Corrected: side effect, dose, product and participant number information in the study table; drug interaction notes; the dose range in the studies; reported side effects. Points that could not be verified from full texts went down from 5 items to 3. A technical phrase ("failed to beat placebo") was replaced with a clearer one ("was not clearly different from placebo"). The year of 4 publications named on the page was corrected to the journal's print year instead of the online publication year. The conclusion and the strength of evidence did not change.
October 19, 2026
Page published. This first review examined 23 studies.

Evidence changes across all topics