Most of what is known about cannabis and autism comes from small trials in which children were given CBD-rich oils for a few weeks. Minnesota has something those trials do not: a law that obliges the state to keep asking its medical cannabis patients how they feel, every time they come back to buy. After six years of collecting those answers from people enrolled for autism spectrum disorder, the state has added them up.
The result is a 39-page report, Patients with Autism Spectrum Disorder in the Minnesota Medical Cannabis Program: Experience of Enrollees During the First Six Years, published on October 1, 2026 by the Division of Medical Cannabis at the Minnesota Office of Cannabis Management (OCM). It is fresh work, and in its press release the agency calls it one of the largest studies on record involving autism and medical cannabis patients. It is also a study without a comparison group, and the authors say so repeatedly. Both things matter for reading the numbers below.
Key Takeaways
Between August 2018 and July 2024, 1,811 people enrolled in Minnesota’s medical cannabis program with autism spectrum disorder as a qualifying condition, and 1,645 of them bought cannabis at least once. Almost nine in ten (88.3 percent) rated their anxiety as moderate to severe when they started. Of those, 57 percent reported at least a 30 percent drop in their anxiety score within four months, and 34.3 percent both reached that drop and held it for at least four more months. The matching figures for lasting improvement were 37.3 percent for disturbed sleep, 34.4 percent for fatigue and 41.3 percent for depression. Children and adults used very different products: families of patients under 18 mostly bought CBD-rich oils and capsules, while adults mostly bought THC-rich products to inhale. About one patient in five reported a side effect, usually a mild one. Because every patient used cannabis and nothing is known about their other treatments, the report cannot show that cannabis caused the improvements, and a behavior questionnaire that looked more directly at autism-related behaviors was completed by only 46 patients.
What the Minnesota Medical Marijuana Autism Study Measured
Autism spectrum disorder became a qualifying condition in Minnesota on August 1, 2018. The report covers everyone who enrolled for it for the first time during the following six years, up to July 31, 2024, with follow-up data collected through July 31, 2025. Of the 1,811 enrollees, 90.8 percent (1,645) made at least one purchase, and the analysis is about them.
The data come from three places: the certification a health care practitioner files when a patient enrolls, the dispensary purchase records, and a form called the patient self-evaluation. During the period studied, every patient had to fill in that form before each purchase. It asks them to rate eight symptoms over the previous 24 hours on a scale from 0 (not present) to 10 (as bad as one can imagine): anxiety, lack of appetite, depression, disturbed sleep, fatigue, nausea, pain and vomiting. A score of 4 or more counts as moderate to severe. The researchers treated a fall of at least 30 percent from the starting score as clinically meaningful, for example from 7 to 4. That threshold is borrowed from clinical trials, where it is best established for pain.
One detail is easy to miss. The eight symptoms are the same for every patient in the program, whatever their diagnosis. They are not a measure of autism itself. The main analysis therefore describes problems that often travel with autism, such as anxiety and poor sleep, and says nothing about social communication or repetitive behavior.
Who the Patients Were
This is a young group. The average age at enrollment was 20, against 45.7 for Minnesota’s medical cannabis patients overall. The youngest patient was 2 years old and the oldest 75. Just under half (47.7 percent) were under 18 when they enrolled.
Almost three quarters (72.3 percent) identified as male and 23.9 percent as female, close to the three-to-one ratio seen across autism research. The report notes that part of that gap reflects how hard autism can be to identify in girls and women. Most patients (77.6 percent) identified as white. About one in five (18.4 percent) had been certified for another qualifying condition as well, most often post-traumatic stress disorder (9.9 percent), chronic or intractable pain (5.0 percent) or seizures (3.2 percent).
Anxiety, Sleep and Depression: The Numbers Behind the Headline
Anxiety dominated. The average anxiety score at enrollment was 7.28 out of 10, and 1,450 patients rated it at 4 or higher. The table shows what happened to the four most common complaints. The last column is the strictest test: the patient had to improve within four months of the first purchase and still show that improvement for at least four months afterward.
| Symptom | Moderate to severe at enrollment | Improved by 30% or more within four months | Improved and held it for four more months |
|---|---|---|---|
| Anxiety | 1,450 (88.3%) | 826 (57.0%) | 498 (34.3%) |
| Disturbed sleep | 1,154 (70.2%) | 700 (60.7%) | 431 (37.3%) |
| Depression | 982 (59.8%) | 623 (63.4%) | 406 (41.3%) |
| Fatigue | 924 (56.2%) | 530 (57.4%) | 318 (34.4%) |
The gap between the two right-hand columns is where most of the story sits. Of the 826 patients whose anxiety improved, only 739 had any data in the following four months, and 498 of those (67.4 percent) were still improved. The report counts everyone else as not having held the gain, including people who simply stopped filling in forms. The authors call this a conservative estimate, and it is: it assumes that anyone who disappeared did not benefit.
Read the other way, the same column says that roughly two in three patients with significant anxiety did not report lasting relief by this measure.
A note on one figure. The running text of the report, and the press release, give 42.9 percent for lasting improvement in depression. The report’s own table and its executive summary give 41.3 percent, which is what 406 out of 982 works out to; 42.9 percent is the table’s figure for lack of appetite. This article uses 41.3 percent.
Children and adults did about equally well. Lasting improvement in anxiety was reported for 33.5 percent of patients under 18 and 35.1 percent of adults; for depression, 41.3 and 41.4 percent; for fatigue, 35.3 and 33.9 percent. Sleep was the exception, with 40.2 percent of adults holding their improvement against 34.1 percent of the younger group.
Children Got CBD-Rich Oils, Adults Bought THC-Rich Products
“Medical cannabis” was not one thing in this study. Purchase records for each patient’s first year add up to 18,102 transactions and 47,071 products, and they split sharply by age.
For patients under 18, a parent or legal guardian has to buy on the patient’s behalf. Those families overwhelmingly chose products that are swallowed: oral solutions, capsules, tablets and gummies made up 85.3 percent of everything bought for this group. Just over half of those products (52.1 percent) contained more CBD than THC, a quarter (25.3 percent) were classed as balanced, and 22.6 percent were THC-dominant.
Adults went the other way. Products for inhalation, meaning vaporizer oils and dried flower, made up 68.0 percent of their purchases. Dried flower has only been sold in the program since March 2022 and only to patients aged 21 and over, yet it accounts for about half of all inhalation products in the data, and 59.0 percent of the flower sold was high in THC and low in CBD.
So the anxiety figures for children and for adults are similar, but they describe two different kinds of treatment. The program also does not record how much patients actually took or how often. The report can say what was bought, not what was used.
The Behavior Questionnaire: 46 Patients Out of 1,170
The researchers did try to look at autism-related behavior more directly. Patients under 26 were invited to complete the Aberrant Behavior Checklist, a standard questionnaire that scores irritability, social withdrawal, repetitive movements, hyperactivity and inappropriate speech, once at enrollment and again about ten weeks after the first purchase. It was optional and had to be mailed back on paper.
Only 46 of the 1,170 eligible patients, or 3.93 percent, returned both forms. In that group, scores fell on all five scales, and each change was statistically significant. The median irritability score dropped from 21.0 to 14.5 and the median hyperactivity score from 27.5 to 19.0. Nearly all the products these patients bought (88.3 percent) were swallowed rather than inhaled, and 54.7 percent were CBD-dominant.
The authors do not lean on this result. The 46 were younger than those who did not respond (12.65 years on average against 15.31) and more often white (91.3 percent against 75.3 percent). Families who saw a benefit were probably more likely to stay in the program and post the second form. The report says plainly that the findings cannot be generalized to the wider group.
Side Effects: One Patient in Five Reported Any
In their first year, 321 patients (19.5 percent) reported at least one side effect, for a total of 1,276 reports. Most were mild (72.6 percent), 21.6 percent were moderate and 5.8 percent were severe, meaning they interrupted usual daily activities. Dry mouth was the most common complaint, at 23.9 percent of all reports, followed by increased appetite, drowsiness and mental clouding.
The 74 severe reports came from 51 patients. The most frequent severe side effect was anxiety (10 reports), the very symptom most patients were hoping to ease, followed by drowsiness (5), fatigue, nausea and panic attacks (4 each).
These figures are probably too low. The form is completed only when a patient returns to buy again, so someone who had a bad reaction and never came back leaves no trace. The report also points out that many patients are young children or have limited communication skills and may not be able to describe what they feel.
Why This Does Not Prove Cannabis Works for Autism
The report’s own conclusion is cautious: patients “were able to achieve and maintain symptom relief after starting medical cannabis”, but because all of them used it, the authors “cannot compare the effectiveness in symptom relief to a control or placebo group”. Several specific problems sit behind that sentence.
There is no way to separate the drug from expectation. Families who enroll a child in a cannabis program hope it will help, and for patients under 18 it is often a parent who fills in the form. In one of the controlled trials the report cites, a study of pharmaceutical CBD in boys with autism, the drug did no better than placebo, and the researchers noted a strong placebo effect.
Nothing is known about other treatments. The analysis assumes cannabis was the only thing that changed. A new medication, a new school or a new therapist during the same months would be counted as a cannabis effect.
The scores come only from people who kept buying. Among all patients, 7.4 percent purchased just once, and 68.5 percent were still purchasing more than eight months after their first visit (73.8 percent of adults and 62.6 percent of those under 18). In earlier surveys, patients who left the program named cost, poor access to dispensaries and lack of effect as their main reasons.
People also tend to seek a new treatment when things are at their worst, and extreme scores usually drift back toward a person’s average on their own. Before-and-after studies cannot rule that out. We saw the same pattern in a 10-year registry of medical cannabis for chronic low back pain, where the improvements were several times larger than in the best randomized trial of the same condition.
How the Findings Compare With Controlled Trials
The picture from placebo-controlled research is mixed, and the report reviews it frankly. An Israeli trial of a 20:1 CBD-to-THC whole-plant extract found better disruptive-behavior ratings than placebo, but no difference on a second scale measuring noncompliant behavior, and a placebo-controlled study of sleep using the same extract found no difference. A Brazilian trial in children aged 5 to 11 reported gains in social interaction, agitation and anxiety after 12 weeks. The trial of pharmaceutical CBD in autistic boys mentioned above found no advantage over placebo. We have covered separately an Israeli placebo-controlled trial of a CBD-rich extract in children and teens with autism.
The closest relatives of the Minnesota report are the UK Medical Cannabis Registry studies, which also follow real patients without a control group. The first, with 74 patients, found better sleep and quality of life over six months; a follow-up published this year reported improvements in anxiety and sleep at 18 months.
For anxiety in general, outside autism, the evidence is still disputed, and a Lancet study has questioned how effective medical cannabis is for it.
Minnesota’s contribution is scale and routine. “There are very few clinical trials using cannabis as a therapeutic agent in patients with autism spectrum disorder, so this report provides important data for future studies on medical cannabis and ASD,” Grace Christensen, an OCM senior research analyst, said in the agency’s release. Daren Howard, interim executive director of the Autism Society of Minnesota, which was consulted on the study, called it “an important step toward safely informing the public and expanding access to supportive treatment options.” The same data system has produced earlier reports on other conditions, including the one behind our article on medical cannabis and sleep apnea in Minnesota, and the full series is listed on OCM’s medical cannabis reports page.
Where the Law Stands in Minnesota and Federally
In Minnesota, autism spectrum disorder that meets the DSM-5 criteria is on the state’s list of qualifying medical conditions, and a health care practitioner must certify the condition before a patient can enroll. The same list now ends with a general clause covering any condition for which the patient’s practitioner has recommended cannabis. According to the program’s key dates, the annual enrollment fee was eliminated on July 1, 2023. Patients under 18 cannot buy for themselves, and dried flower is restricted to those 21 and older.
Federally, an order published in the Federal Register on April 28, 2026 moved two categories to Schedule III of the Controlled Substances Act: FDA-approved drug products containing marijuana, and marijuana covered by a state medical marijuana license. Everything else stayed in Schedule I. That does not make dispensary products approved medicines, and no cannabis product is FDA-approved for autism. As we reported earlier, rescheduling by itself does not remove the main obstacles to cannabis research, which is one reason state reports like this one fill a gap that trials have not.
FAQ
Among 1,645 patients enrolled for autism spectrum disorder who bought medical cannabis, 88.3 percent started with moderate to severe anxiety. Of those, 57 percent reported at least a 30 percent reduction within four months, and 34.3 percent reached that reduction and kept it for at least four more months. Lasting improvement was reported by 37.3 percent of patients with disturbed sleep, 34.4 percent with fatigue and 41.3 percent with depression.
No. The main analysis measured eight general symptoms such as anxiety and sleep, not the core features of autism. There was no control or placebo group, the researchers had no information on other treatments, and the ratings came only from patients who kept returning to buy. A questionnaire on autism-related behavior was completed by 46 patients, too few and too self-selected to draw conclusions from. The authors themselves say more research is needed.
In Minnesota, yes: autism spectrum disorder has been a qualifying condition since August 1, 2018, and enrollment requires certification by a health care practitioner. Patients under 18 need a parent or legal guardian to purchase for them, and dried flower is limited to patients 21 and older. Other states set their own lists of qualifying conditions, and not all of them include autism. At the federal level, marijuana covered by a state medical license was moved to Schedule III in April 2026, while other marijuana remains in Schedule I. No cannabis product has been approved by the FDA for autism.
About one in five patients (19.5 percent) reported a side effect during their first year. Of the 1,276 reports, 72.6 percent were mild, 21.6 percent moderate and 5.8 percent severe. Dry mouth was the most common. Anxiety was the most common severe side effect. The authors consider these figures an undercount, because patients who stopped buying after a bad reaction were not recorded.
Legal Disclaimer
This article is for informational and educational purposes only. It summarizes a report published by a state agency and does not constitute medical or legal advice. Nothing here is a recommendation to use, buy or give cannabis or any cannabinoid product to a child or an adult, or guidance on qualifying for a medical cannabis program. The products described are not approved by the U.S. Food and Drug Administration for autism spectrum disorder. Cannabis laws differ from state to state and between state and federal level, and they change often. Decisions about treatment, especially for children, should be made with a licensed health professional who knows the patient; questions about the law in your jurisdiction should go to a qualified attorney or the relevant state agency.
