Medical cannabis collection · Sources reviewed 11 October 2026
CBD is being investigated for psychosis, but an investigational treatment is not an established treatment. Some small trials suggest benefit when CBD is added to usual care. Others do not. Two systematic reviews published in 2026 illustrate how a promising signal can remain uncertain, even after researchers combine the available studies.
On this page

First separate three different questions
Researchers may ask whether CBD helps someone already experiencing psychosis, whether it prevents psychosis developing, or whether it protects against unwanted effects of THC. These are different clinical questions. A favourable result in one does not establish the others.
CBD and THC are also different substances. A trial of purified CBD does not validate cannabis flower or an oil containing an uncertain mixture. The formulation and accompanying treatments are part of what the trial tested.
The review that found a small signal
Di Francesco and colleagues pooled five randomised trials comparing additional CBD with placebo in people with schizophrenia spectrum disorders. Their 2026 review found small statistically significant improvements in total, positive and general symptom scores, but no significant improvement in negative symptoms. The total-score difference was about two points on the Positive and Negative Syndrome Scale, or PANSS. The adjunctive-treatment review.
In this terminology, positive symptoms are experiences such as hallucinations or delusions; negative symptoms include diminished motivation or emotional expression. The words do not mean good and bad. A small statistical difference on a symptom scale also needs to be considered alongside its practical effect on distress, relationships and everyday life.
The broader review did not find a clear benefit
Marchi and colleagues included eight trials, six published and two unpublished, involving 288 participants. Their search extended through July 2025, and the included studies ranged from brief experiments to 12-week treatment. Comparators included placebo and amisulpride. The pooled result did not establish a clear improvement in psychosis symptoms, and the authors rated the evidence as low or very low certainty. The broader 2026 review.
“CBD did not show a clear benefit for psychosis symptoms in RCTs”
Marchi and colleagues, 2026RCTs are randomised controlled trials. The authors describe uncertainty in the current evidence, rather than proving that every possible CBD regimen is ineffective.
How can both reviews exist?
They did not include precisely the same evidence or define the treatment question in the same way. One focused on CBD added to treatment versus placebo; the other admitted a broader set of designs and comparators, including unpublished results. Search dates and analytical choices also differed.
These are plausible reasons for different findings, not a claim that any single choice explains the discrepancy. Establishing that would require examining and reanalysing the individual datasets. Because the reviews overlap, they should not be counted as independent sets of patients providing separate confirmation.
For readers, the useful response is to compare eligibility criteria, the size of the effect and its uncertainty. Selecting whichever review matches a preferred answer would conceal the central finding: the evidence remains fragile.
Does CBD protect against THC?
A 2025 double-blind crossover experiment studied 30 people who had schizophrenia or schizoaffective disorder alongside cannabis use disorder. Participants received CBD or placebo before a controlled THC exposure. CBD did not protect memory or psychotic symptoms; delayed recall and the increase in positive symptoms were worse under CBD pretreatment in this experiment. The THC-challenge trial.
This was a small, acute study in a specific population. It does not show that CBD worsens schizophrenia in every setting. It does challenge the reassuring assumption that adding CBD necessarily makes THC safe for someone vulnerable to psychosis. The controlled exposure used in research is not an invitation to reproduce the experiment.
Questions worth resolving
These questions leave room for further research without making a current treatment promise. Someone considering CBD should discuss the actual product and their existing medicines with their treating team. The published studies do not support independently stopping or replacing antipsychotic medication.
References and further reading
- Di Francesco et al. (2026). Meta-analysis of adjunctive cannabidiol in schizophrenia spectrum disorders.
- Marchi et al. (2026). Cannabidiol for psychotic disorders: systematic review and meta-analysis of randomised trials.
- Randomised THC-challenge trial of CBD pretreatment in schizophrenia and co-occurring cannabis use disorder (2025).
Educational information, not an individual prescribing plan. Decisions about medicines belong with the treating clinician. External reference links open in a new tab.