CBD clinical trials are still a fairly short list, but a clear picture is emerging condition by condition. A useful early review, from Larsen and Shahinas in 2020, covered 25 human studies and 927 participants published through 2019 (not all patients — some studies used healthy volunteers) — but it excluded pediatric studies and studies combining CBD with other cannabinoids, and several conditions have moved on since. Here’s what the trials actually show, condition by condition, including the newer evidence. This isn’t every CBD trial ever conducted; it covers the conditions with the strongest or most frequently cited clinical evidence.
What the trials found, by condition
Epilepsy is the clear exception to nearly everything else here, so it goes first. Randomized trials of purified, pharmaceutical-grade CBD (Epidiolex) reduced seizures in Lennox-Gastaut syndrome and Dravet syndrome, two severe childhood-onset epilepsies — see our original coverage of that approval. The FDA has since approved it for tuberous sclerosis complex too, per the current prescribing label. This is the strongest human evidence for CBD of any condition here — but it’s purified CBD at high, weight-based doses (10–25 mg/kg/day) under medical supervision, not a retail oil. Much of the pivotal epilepsy evidence fell outside Larsen and Shahinas’s adult-only scope because the trials enrolled children or mixed pediatric-and-adult populations. Full detail in our epilepsy article.
Psychosis and schizophrenia are genuinely unsettled, and two 2026 meta-analyses disagree. One pooled 8 RCTs (288 patients) and found no significant benefit anywhere — symptoms, cognition, or subscales — though it only had power to rule out a fairly large effect. A second, restricted to 5 trials of CBD added to existing antipsychotics (204 patients), found small but significant improvement in total, positive, and general symptoms (not negative symptoms). The honest summary: a possible small adjunctive benefit, still uncertain — not that the earlier positive signal was wrong.
Anxiety has real new evidence, with an asterisk. Single 300–600 mg doses helped in older, mostly single-dose studies, but a 2024 review of 15 RCTs found single doses often failed in more recent trials, and higher doses sometimes made things worse. The strongest positive result is a 2024 phase 3 trial: 178 patients took daily oral CBD for 15 weeks and improved significantly more than placebo on two major anxiety scales (p<0.0001 both). But several of that trial’s authors were employees or executives of the company developing the formulation, and it hasn’t been independently replicated — and a 2026 Lancet Psychiatry review of 54 trials found no significant overall anxiety benefit from cannabinoids generally. Fair takeaway: sustained daily dosing may be more promising than the inconsistent single-dose evidence, but one positive, unreplicated trial does not establish that pattern.
Pain is genuinely mixed. A 2024 review found 11 small and highly varied human studies, with mixed results across different products and pain conditions. One encouraging example: a 2026 trial of high-dose CBD (up to 800 mg/day) in about 40 patients with neuropathic pain from spinal cord injury found a statistically significant benefit — though the average advantage over placebo was modest, despite a stronger result on the share of patients who got meaningful relief (37.8% versus 11.1%). A less encouraging example: a well-powered 2023 trial of 600 mg/day CBD added to paracetamol for knee osteoarthritis found no benefit over placebo at all — and notably, more liver-enzyme elevations on CBD. There’s no single answer for “does CBD help pain”; it depends heavily on which pain and which trial.
Substance use was mixed and mostly underwhelming: CBD isolate (200–800 mg) didn’t reduce cannabis’s reinforcing effects in cannabis smokers, and inhaled CBD (400 mcg) modestly cut cigarette counts in smokers without reducing cravings.
Diabetes, Crohn’s, and ulcerative colitis are each more nuanced than “no benefit.” Diabetes: 200 mg/day for 13 weeks did not improve fasting glucose, HbA1c, or the study’s main cholesterol outcome; a different cannabinoid tested in the same trial, THCV, produced the more encouraging findings, significantly lowering fasting glucose (our diabetes article covers more encouraging newer evidence on blood sugar and nerve pain specifically for CBD). Crohn’s: pure CBD (20 mg/day) failed outright, but a later trial of a 4:1 CBD-to-THC oil found improved symptoms and quality of life without improved inflammation on endoscopy — a real but modest symptomatic signal, not disease treatment. Ulcerative colitis: a similar CBD-rich, THC-containing extract missed its main remission target (28% vs. 26% placebo) over 10 weeks, but because many patients couldn’t tolerate the full dose, some per-protocol symptom and patient-reported measures favored the extract in those who could — a negative headline result with a real asterisk, not a flat “no benefit.”
Side effects and safety
CBD was generally well tolerated across these trials — mild to moderate effects, mainly tiredness, diarrhea, and appetite or weight changes. At the high, prescription-strength doses used for epilepsy, there’s a specific concern: dose-related liver-enzyme elevations, seen in about 12–13% of Epidiolex patients versus 1% on placebo — rising to 21% in patients also taking valproate, and to 30% in patients taking both valproate and clobazam together — and the osteoarthritis trial above saw a similar signal even at a lower 600 mg/day dose. Our drug-interactions guide and doctor-questions checklist cover this in more depth.
Why the evidence base is still thin
Regulation is one reason this list is still short, but not the only one — trial cost, limited funding, inconsistent formulations, and the difficulty of standardizing retail extracts all play a part too. Federal scheduling historically added cost and delay to cannabis research. The 2018 Farm Bill removed hemp-derived CBD from Schedule I, and in April 2026 certain FDA-approved and state-licensed medical cannabis moved to Schedule III (our legal-status guide has the current state-by-state picture). Other marijuana remains under stricter federal controls while broader rescheduling is considered, so the research system is still uneven. Isolate is also scientifically easier to study, regulation aside: researchers know which single compound produced a given result, which is part of why most of the trials above used CBD isolate rather than full-spectrum extracts. Larsen and Shahinas were candid about the field’s limits: larger, more diverse samples, more standardized outcome measures, and longer-term safety data are what future trials still need.
The bell curve and the “entourage effect” — what’s actually established
Two claims recur in CBD marketing and both need correcting. The “bell-shaped” dose response — working best at a middle dose — started in 2015 mouse research by Gallily, Yekhtin and Hanuš, but two small human public-speaking studies found a similar pattern: 300 mg reduced anxiety while 100 mg and 900 mg didn’t. That’s evidence for the pattern in one acute, anxiety-specific setting — not proof of one ideal dose for every condition, and it says nothing about full-spectrum products specifically.
The related claim — that full-spectrum products dodge this problem via an “entourage effect” from terpenes and other cannabinoids — is a hypothesis, not a fact. Human evidence is thin: a recent comprehensive review found most promising terpene interactions from lab studies simply don’t replicate in controlled human trials. Whether full-spectrum CBD behaves meaningfully differently from isolate in people remains genuinely open.
This isn’t medical advice. Talk to a doctor or pharmacist before using CBD for a specific condition, especially alongside other medications — see our related pieces on CBD and schizophrenia, CBD and epilepsy, and full-spectrum vs. broad-spectrum vs. isolate.
