Cannabis Edibles and Schizophrenia

A familiar pitch on high-CBD and 1:1 products is balance: take it with your THC and it takes the edge off. In a trial published in 2025, a London research team tested that promise in the people who most need it to be true. Thirty adults with schizophrenia or schizoaffective disorder and a cannabis use disorder took 1,000 mg of oral CBD or a placebo, then inhaled cannabis. After CBD, their memory was worse and their psychotic symptoms climbed higher.

This page is for people living with a schizophrenia diagnosis and the families reading alongside them; who develops psychosis in the first place is on the psychosis risk page. The position is short. THC is off the table after diagnosis. CBD at gummy doses has never been tested for the illness, and at trial doses it did little. The change with the clearest consequence is the one that sounds healthiest: swapping joints for edibles while taking clozapine.

Does CBD protect against THC in people with schizophrenia?

No. In the first trial to test it in this population, 1,000 mg of CBD taken three hours before cannabis left people with schizophrenia recalling fewer words and with a bigger rise in positive psychotic symptoms than placebo did. The authors concluded their findings do not support CBD as a way to soften cannabis's acute effects.

Chesney and colleagues ran it at King's College Hospital between 2021 and 2023. Everyone took a regular antipsychotic, and 27 of the 30 had severe cannabis use disorder. Each did a pair of visits at least a week apart, swallowing five 200 mg CBD capsules on one and identical placebo on the other, then inhaling vaporised flower carrying 20 mg of THC three hours later. Of the 11 whose symptoms barely moved at 20 mg, six came back for a pair at 40 mg, and one of those for a third pair at 60.

Delayed recall of a 12-word list averaged 3.5 words after CBD and 4.8 after placebo. Positive symptoms on the PANSS rose 5.0 points with CBD against 2.9 with placebo, and seven people jumped nine points or more after CBD, none after placebo. Both treatment-related adverse events followed CBD. One was a bout of persecutory delusions about the study team, serious enough to withdraw that participant, which cleared by the next day.

CBD did not raise blood levels of THC or its active metabolite, so the authors rule out a blood-level interaction and suspect CBD changed how the body responded to THC. The catch for an edibles site is that this cannabis was inhaled. The same paper notes that taken together by mouth, CBD can slow the liver's first-pass breakdown of THC and strengthen it. A 1:1 gummy is the swallowed version.

The balance claim has now been tested once in people with schizophrenia, at forty times the CBD in a 25 mg gummy, and it ran in reverse. No THC:CBD ratio clears the first test in our medical edible standard for this reader. UCLA researcher Anya Bershad's commentary on the trial put the clinical message plainly: patients should hear that CBD "probably won't protect them from cannabis-related harms." The sample was 28 men and two women.

Can CBD treat schizophrenia?

No trial has tested anything close to a gummy dose. At 800 and 1,000 mg a day of pharmaceutical CBD, two trials reported improvement; four at 300 to 1,000 mg found none. The one positive placebo-controlled result, McGuire's 2018 trial, came to 1.4 points on a symptom subscale that runs from 7 to 49.

Leweke's 2012 trial in Cologne randomised 42 inpatients with acute paranoid schizophrenia to CBD or the antipsychotic amisulpride, each stepped up to 800 mg a day, with no other antipsychotic, for four weeks. Symptoms fell about 30 PANSS points in both groups. CBD came with fewer movement side effects, less weight gain and a smaller prolactin rise, and a rise in anandamide, a cannabinoid the body makes, tracked the improvement. Funding ran out at 21 patients per arm against 35 planned, and the formal non-inferiority test did not pass.

McGuire's is the largest placebo-controlled trial: 88 patients at 15 hospitals in the UK, Poland and Romania added 1,000 mg a day of CBD oral solution, or placebo, to their usual antipsychotic for six weeks. Positive symptoms fell 1.4 points more on CBD, and clinicians rated 78.6% of the CBD group improved against 54.6%. Negative symptoms, total PANSS, cognition and functioning did not differ significantly, and the authors called the effect modest. Adverse event rates matched placebo overall, with stomach complaints higher on CBD (nine patients against three). GW Research funded it and supplied the drug, and half the authors worked for GW.

The trials since found nothing. Boggs's Yale group (600 mg, six weeks, 36 stable outpatients) saw no benefit on symptoms or cognition, and van Boxel's Utrecht study (600 mg, four weeks, 31 people within five years of diagnosis) saw no clinical effect. Jazz Pharmaceuticals, which now sells Epidiolex, tested 300 and 1,000 mg against placebo for 12 weeks in 77 people with an inadequate response to their antipsychotic, and neither dose beat placebo on any symptom measure. Jazz terminated the study for a business decision and posted the results to ClinicalTrials.gov in June 2023. No journal paper has followed, and a small Yale crossover trial at 800 mg sits unpublished on the same registry (see the table).

The six controlled treatment trials, side by side:

How many CBD gummies would it take to match the trials?

Forty, at 25 mg a gummy. McGuire's positive result came from 1,000 mg of CBD a day, so matching the trial with the 25 mg gummy priced below means forty every day for six weeks. At namaCBD's best one-time price, that is $59.33 a day for a benefit measured at 1.4 points.

The gummy is namaCBD Relax, the THC-free option on our perimenopause anxiety page: 25 mg of broad-spectrum CBD per piece, $34 for 20 or $89 for 60 on the brand's site as of September 29, 2026. Forty a day is $59.33 at the 60-pack price, about $2,490 for six weeks, against brand directions of one or two. Each gummy also carries 100 mg of ashwagandha, so the trial dose by candy adds four grams of an herb nobody in the trial took. Chesney's participants swallowed their 1,000 mg in five capsules.

The arithmetic is here to price the trial dose, and nobody should read it as a regimen.

Why do the 2026 reviews disagree?

Because they pooled different trials and measured different things. Di Francesco's positive-symptom estimate pooled four placebo-controlled add-on trials, where McGuire carries more than half the weight, and found a small benefit. Marchi's and Wilson's wider pools, which reach into single-dose experiments, found no significant effect. Either way the effect is small.

Di Francesco's group pooled Boggs, McGuire, van Boxel and the Jazz 1,000 mg arm, 191 people. Positive symptoms came out 1.30 points lower on CBD, negative symptoms showed no significant difference from placebo, and the authors called the advantage "statistically significant, albeit small." McGuire supplies 55.9% of that estimate. Checked against the primary sources, the review also gets two inputs wrong: McGuire's paper reports six weeks of treatment where the review lists eight, and the Jazz registry shows 95 people entering a placebo run-in, 77 randomised and a 300 mg arm, where the review describes a 95-person trial of 1,000 mg.

Marchi's group pooled eight trials and 288 people, two of the trials unpublished, at a median 800 mg a day, and found no significant effect on overall, positive or negative symptoms. Wilson's Lancet Psychiatry review of 54 cannabinoid trials in mental and substance use disorders had eight in psychotic disorders, 290 people: seven CBD trials and one intravenous THC experiment. No PANSS result reached significance.

The three reviews, and what each one pooled:

How many people with schizophrenia have a cannabis use disorder?

About one in four has had a cannabis use disorder at some point, and about one in six has one now. Koskinen's 2010 meta-analysis of clinical samples put the median lifetime rate at 27.1% across 28 studies and the current rate at 16.0% across 10. Younger samples ran much higher.

In samples averaging under 30, the median current rate was 38.5%. The studies date from 1996 to 2008, cover people already in treatment, and the authors flag that illegality pushes self-report down. This is the group Chesney's trial was run in.

What happens if cannabis use continues after diagnosis?

More relapse. Schoeler's 2016 meta-analysis of 24 studies and 16,565 people with psychosis found that continued users relapsed more than non-users and more than people who stopped, and spent longer in hospital than non-users. People who stopped showed no difference in relapse from people who had never used cannabis.

Against non-users, the relapse effect of continued use was 0.36 standard deviations; against people who quit, 0.28. In the seven studies that counted relapse as yes or no, continued users had 1.97 times the odds of relapse that non-users had. The authors compare it to the relapse effect of high expressed emotion in a household (0.31). The pool covers psychosis broadly, schizoaffective and bipolar illness included, and it is observational: the authors were unable to rule out that people who manage to stop are a milder group to begin with.

Why does switching from joints to edibles matter on clozapine?

Because smoke speeds up the liver enzyme that clears clozapine and olanzapine, and edibles contain no smoke. Tobacco and cannabis smoke both induce CYP1A2, and the effects add up. When the smoke stops, the enzyme slows within days, so the same prescribed dose produces higher blood levels. A 2002 case report describes clozapine intoxication after exactly that.

Anderson and Chan's 2016 review in Clinical Pharmacokinetics lays out the mechanism: combustion products switch on CYP1A2, marijuana smoke does it as well as tobacco, and "the induction effect between the two products is additive." The review puts clozapine and olanzapine among the drugs smokers clear 30 to 100% faster. Chlorpromazine, its one antipsychotic with direct cannabis data, cleared 107% faster in people who smoked both. The authors infer that combustion products do most of the inducing; nicotine without smoke leaves the enzyme alone.

The reversal is quick. In heavy tobacco smokers who quit, caffeine clearance (the standard CYP1A2 readout) fell 12.3% on day one and 28.2% by day four, and the review describes a clinically significant effect within a week and empirical dose reductions within two to three days.

Zullino and colleagues reported the clinical version in 2002: a patient on clozapine who stopped smoking tobacco and cannabis became confused as clozapine levels rose, and a second, on olanzapine, developed marked movement side effects after cutting back on tobacco. Their conclusion was that smoking on these drugs should be monitored so doses can follow it. A switch to edibles is smoking going down. The THC keeps coming and the smoke stops, so from the enzyme's side the switch is a quit attempt. Our medical use guide makes the general case for telling a doctor about edibles; on clozapine it comes with a timetable.

Does CBD change clozapine levels?

The evidence so far comes from caffeine. CBD inhibits CYP1A2, the enzyme that clears clozapine: the Epidiolex label says "Cannabidiol is a weak inhibitor of CYP1A2," and in Thai's 2021 trial, 750 mg of CBD twice a day nearly doubled healthy volunteers' exposure to caffeine, the standard probe for that enzyme.

Caffeine's peak rose 15% and its total exposure 95%, figures the label repeats. Neither source names clozapine or olanzapine (the label's examples are theophylline and tizanidine), both come from 1,500 mg a day, and nobody has tested a gummy dose against the enzyme. The same trial withdrew six of its 16 volunteers because their liver enzymes rose, and six in all had rises of at least three times the upper limit of normal.

Where is CBD research in schizophrenia going next?

To people for whom clozapine has not worked. CanCloz, a Queensland trial of 1,000 mg of CBD a day added to clozapine for 12 weeks, is the first test of CBD in clozapine-resistant schizophrenia. Its registry entry says it stopped early for lack of funding, staff or facilities, at 60 of a planned 88 participants, and that the collected data are being analysed. No results are public yet.

Its protocol measures clozapine levels before and after, the number the section above is missing. It will not answer the balance question: CanCloz excludes current cannabis users, Leweke turned away 79 of 198 people screened for cannabinoids in their urine, and McGuire had three THC-positive patients out of 88, too few to analyse. The only controlled CBD study built around people with schizophrenia and a cannabis use disorder, roughly one patient in six, is Chesney's, and it measured symptoms brought on by THC. Whether CBD treats the illness in that group has never been tested.