Search for cannabis and autism and nearly everything that comes back is about children. That is because nearly all of the research is about children: Israeli, Brazilian, American and Australian trials of CBD-rich oils given to kids, most of them with severe behaviour problems, scored by their parents. An autistic adult reading those results for themselves is working from a population that does not include them.
Our position, after reading every trial we could verify: the controlled evidence sits in children, most of those trials missed at least one of their main targets, and the products tested were prescription-grade oils dosed by body weight three times a day. An adult buying a gummy is extrapolating twice, once across age and once across product. That leaves a decision autistic adults are entitled to make for themselves, and this page sets out what the evidence can and cannot tell them about it.
Does CBD help autism?
In children, the controlled trials are mixed at best, and in adults there are none. The largest trial, 150 young people in Jerusalem, hit one co-primary outcome and missed the other. A 2026 San Diego trial of Epidiolex found CBD did no better than placebo on its main scales, and an Australian crossover missed its primary outcome too.
The Jerusalem trial, Aran and colleagues in Molecular Autism in 2021, is the one every CBD brand quietly leans on, so it is worth reading closely. Participants were 5 to 21 years old with a mean age of 11.8, and 80% were boys. On the parent-rated Home Situations Questionnaire, the registered primary outcome, nothing separated the groups. On the clinician-rated global impression of behaviour problems, the co-primary, 49% improved on the whole-plant extract against 21% on placebo. The authors called the evidence for efficacy "mixed and insufficient," which is a more honest summary than most product pages manage. Aran disclosed advisory fees and stock options from BOL Pharma, which supplied the cannabis and part-funded the study. A sleep analysis from the same trial (Schnapp 2022) found the cannabinoids no better than placebo on any sleep measure, including how long children took to fall asleep.
The San Diego trial (Trauner 2026) is the cleanest test of pure CBD: autistic boys aged 7 to 14, up to 20mg/kg a day of Epidiolex, eight weeks against placebo. Both arms improved on the repetitive-behaviour and behaviour-checklist scores, with no significant difference between them. The authors flagged a strong placebo effect and noted that the behaviour medications the boys already took appeared to lower their CBD blood levels. The Australian crossover (Parrella 2026, 29 children aged 5 to 12) missed its primary social responsiveness outcome and picked up secondary signals on anxiety and parental stress.
The positive Brazilian trial, Silva Junior and colleagues, randomised 60 children aged 5 to 11 and reported gains in social interaction, anxiety and agitation. It is the strongest result in the set, and its concentration gain held only in children with mild autism. The 2026 Lancet Psychiatry meta-analysis that pooled cannabinoid trials across mental disorders found a small reduction in autistic traits (a standardised mean difference of 0.36) and rated the evidence quality generally low, and its authors concluded that routine cannabinoid treatment for these conditions is rarely justified.
What did the trials actually give people?
Prescription-grade oils, dosed by weight and split across the day. In the Jerusalem trial, the extract ran 20 parts CBD to 1 part THC, titrated every other day to a ceiling of 420mg of CBD and 21mg of THC daily, in three doses. Nobody in a controlled autism trial was handed a 25mg gummy.
That 420mg figure deserves a moment at the dispensary counter. At 25mg of CBD per gummy, that ceiling works out to nearly 17 gummies a day, which no label on the shelf suggests and no wallet enjoys. The ratio is the other gap: at 20:1, the trial oils carried 1mg of THC for every 20mg of CBD, the reverse of the THC-forward gummy most dispensaries sell. The delivery differs too. A gummy goes through the gut on its own schedule, with onset running 30 minutes to 2 hours according to our edible timing guide, while the trial solutions were given three times a day.
One more fact frames all of this. The outcome these trials chased, irritability and aggression in autistic children, already has two FDA-approved drugs. The risperidone label lists irritability associated with autistic disorder, established in trials of children aged 5 to 17, and the aripiprazole label carries the same indication for ages 6 to 17. The cannabis trials were competing with antipsychotics, on a paediatric outcome, from the start.
What do autistic adults report about cannabis?
Relief, in the short term, with no control group to check it against. A 2025 analysis of 111 autistic adults tracking sessions in a symptom app found self-rated severity fell by about 73% after use. A UK registry of 130 autistic adults on prescribed cannabis reported better anxiety and sleep scores over 18 months, also without a comparison arm.
The app study (Karhson, LaFrance and Cuttler, in Scientific Reports) is the first look at acute effects in autistic adults, and its fine print matters for this site in particular. The authors kept only inhaled sessions, because oral and inhaled cannabis start working on different clocks, so edibles are absent from the data. They also found that baseline repetitive-behaviour and negative-affect ratings rose over time as people kept using, and offered two readings: regular use making those symptoms worse, or people reaching for cannabis at higher symptom levels as time went on. Diagnoses were self-identified.
The UK registry paper (Aggarwal 2026) comes from a team most of whom list Curaleaf Clinic in London as an affiliation, and to its credit says plainly that its results "represent associations rather than proven treatment effects." Twenty-five of the 130 patients reported 232 adverse events between them.
The one placebo-controlled adult study is a brain scan. Pretzsch and colleagues at King's College London gave 17 autistic and 17 non-autistic men a single 600mg dose of CBD and measured glutamate and GABA two hours later. CBD shifted GABA in opposite directions in the two groups, and the authors wrote that their results do not speak to whether CBD works.
Is it the autism or the anxiety and sleep?
Mostly the co-occurring conditions, on the available evidence. A 2019 Lancet Psychiatry meta-analysis put ADHD at 28% of autistic people, anxiety disorders at 20% and sleep-wake disorders at 13%. In autistic adults specifically, current anxiety ran at 27% and lifetime anxiety at 42%, per a separate 2019 review.
The UK survey by Hua and colleagues at Bath points the same way. Autistic adults were about 3.5 times as likely as matched controls to have used CBD in the past year, and cannabis use did not differ. Frequency of CBD use did not track autism-related difficulties, and the authors suggested the higher rates of anxiety and depression in the autistic group as the likelier driver. Each of those conditions has its own evidence base, much of it in adults, on our pages for generalized anxiety, sleep onset and ADHD. Reading those is more useful than reading the paediatric autism trials, because they at least studied adults.
Is Epidiolex approved for autism?
No. The Epidiolex label lists seizures associated with Lennox-Gastaut syndrome, Dravet syndrome and tuberous sclerosis complex, in patients 1 year and older. The San Diego trial above used Epidiolex and missed on behaviour. Epilepsy is common alongside autism, and in that case the drug is prescribed for the seizures.
The 2020 systematic review by Fusar-Poli and colleagues observed that cannabinoids cut seizure frequency in autistic patients with epilepsy and let some reduce other medications, while knowledge of effects on core autism features was scarce. Our epilepsy page covers the seizure evidence, which is far stronger than anything on this page.
What should the prescriber know?
Everything, and especially about THC. The risperidone label advises caution with other centrally acting drugs and warns of drowsiness and blood pressure drops on standing. The aripiprazole label warns of cognitive and motor impairment. The Epidiolex label says CBD with other central nervous system depressants raises the risk of sedation.
None of those labels names cannabis, and no study has tested a THC edible with either antipsychotic, which is the reason the prescriber needs to hear about it first. For stimulants, our ADHD page covers the one lab study of oral THC with methylphenidate, where the heart-rate effects stacked. For SSRIs we found no autism-specific data at all. The psychiatric risk with THC is real in this population too: in the 2019 Jerusalem retrospective of 60 children on CBD-rich cannabis, one girl on a higher-THC product had a transient serious psychotic event that needed an antipsychotic. Our psychosis risk page covers who carries that risk and why.
Getting to the conversation at all is part of the problem. In the Bath survey, autistic adults were far more likely than controls to name barriers to cannabinoid-related support, with "being in a crowded or chaotic place" carrying an odds ratio of 9.79 and "going somewhere unfamiliar" 5.29. The same group trusted doctors less as a source of cannabis information. A busy dispensary on a Friday evening is, by that measure, close to the worst-designed place to get dosing advice, and the person behind the counter is not the one holding the medication list. The Bath authors read their results as a guide to how cannabinoid information should reach autistic adults.