Man sleeping. (Courtesy of Shane on Unsplash)
A Cannabis-Based Pill Just Cut PTSD Nightmares in a Major Clinical Trial
In A Nutshell
- A large clinical trial found that dronabinol, a THC-based pill, significantly reduced the frequency and intensity of PTSD-related nightmares compared to a placebo.
- More than a third of patients on the drug saw their nightmares disappear entirely, versus 15 percent on placebo.
- Side effects were more common with dronabinol, including seven serious adverse events, but researchers found no signs of withdrawal after stopping treatment.
- The ten-week trial cannot say whether benefits last long-term or whether dependence could develop with extended use.
A pill made from the main active compound in marijuana just succeeded where years of other treatments have fallen short. In a rigorous clinical trial, the drug cut the frequency and intensity of nightmares tied to post-traumatic stress disorder, with no clear withdrawal signal after treatment ended, a reassuring sign for a cannabis-based psychiatric medication that might otherwise be a hard sell.
For trauma survivors, evenings and attaining a good night’s sleep can be especially difficult. Nightmares affect roughly half to seventy percent of people with PTSD and are linked to substance abuse, suicidal thoughts and suicide attempts. Yet doctors have almost nothing specifically designed to stop them. Antidepressants ease some symptoms but do little for nightmares. Prazosin, a blood pressure drug long used off-label for trauma nightmares, looked promising in small trials, then failed in a large one. Against that backdrop of dead ends, a treatment that actually works stands out.
A study published in Nature Medicine offers the strongest evidence yet that a cannabis-derived medication can help. Researchers ran a multicenter trial at four German medical centers, testing dronabinol, a pharmaceutical form of THC, against a placebo in 171 adults with PTSD and frequent nightmares. Patients taking dronabinol saw significantly greater relief in both frequency and intensity of nightmares than those on placebo, along with better sleep and fewer self-reported PTSD symptoms. It is not a cure, but the results mark a meaningful step forward.
Trial Enrolled Mostly Female Civilian Survivors With PTSD Nightmares
Doctors at four German university medical centers enrolled patients between October 2020 and January 2025 for the trial, formally named THC-PTSD, though each participant was treated for only ten weeks. Participants had diagnosed PTSD and nightmares at least twice a week, frequent and intense enough to matter clinically. The average participant was around 38 years old, and roughly four out of five were women. Most had survived physical or sexual assault, often beginning in childhood or adolescence, and only a handful had combat or war-zone trauma. Because the trial population was mostly female civilian survivors of violence, the results may not translate as cleanly to combat veterans or other groups with different trauma histories.
Participants split roughly in half: 87 received dronabinol, 84 received a placebo designed to look, taste and smell identical, down to a cannabis flavoring so taste alone would not give away the assignment. Dosing started low, at 2.5 milligrams, and rose over four weeks to a personalized dose as high as 15 milligrams, taken once daily before bed, followed by six weeks at a steady dose, for ten weeks total.
More Than a Third of Patients on Dronabinol Saw Nightmares Disappear
Researchers measured nightmares using a standardized clinician-administered PTSD assessment rather than relying only on a simple patient rating. After ten weeks, patients on dronabinol saw their nightmare scores drop nearly twice as much as those on placebo. More than half of dronabinol patients had their nightmares cut by half or more, compared to under a third on placebo. Even more telling, over a third saw their nightmares disappear completely, versus 15 percent on placebo.
Benefits extended beyond nightmares. Patients on dronabinol reported better overall sleep and felt their PTSD symptoms had eased. Asked how much better they felt overall, they were far more likely to say meaningfully so. Curiously, the drug did not clearly outperform placebo when doctors scored overall PTSD severity, suggesting the benefit centered on sleep and nightmares rather than every trauma symptom.
Side Effects and an Imperfect Blind Complicate the Picture
Safety was a mixed picture. Side effects were common in both groups but more so with dronabinol: nearly 90 percent had some adverse event, versus 77 percent on placebo, though most were mild or moderate, such as dry mouth, dizziness, tiredness or nausea. Seven dronabinol patients had a serious adverse event, including two overdoses involving the study medication, versus zero on placebo, though researchers concluded none were caused by the drug itself. No one died, and roughly six percent in each group quit due to side effects. A standard withdrawal checklist found no meaningful increase in withdrawal symptoms after stopping, though the short follow-up cannot rule out dependence developing with longer use. One caveat worth noting: because THC’s effects are hard to hide, many patients could likely tell whether they got the real drug, which means some of the reported improvement may be tangled up with knowing, or suspecting, they were on the real thing.
Dronabinol will not replace therapy, and it is not a fix for PTSD as a whole. But for a symptom that has quietly wrecked sleep, worsened depression and, in the worst cases, contributed to suicide, a drug that cuts nightmares in a rigorous, placebo-controlled trial matters. Serious side effects turned up only in the dronabinol group, seven cases against none on placebo, and investigators judged none connected to the drug, but a trial this size and length cannot settle questions about rare or long-term risks. The bigger test now is durability: whether relief holds up over months and years, and whether the safety tradeoffs remain acceptable outside a closely monitored ten-week study.
Disclaimer: This article is for informational purposes and does not constitute medical advice. Dronabinol is a prescription medication, and any decisions about treatment for PTSD or related conditions should be made in consultation with a licensed healthcare provider.
Paper Notes
Limitations
The trial ran for only ten weeks, so it cannot say whether dronabinol’s benefits persist over the long term, whether tolerance develops, or whether dependence emerges with continued use beyond what a short withdrawal checklist could detect. The sample size, while adequate to detect the main effect, was too small to reliably catch rare side effects or draw firm conclusions about how the drug performs in specific subgroups, including by sex, since men made up only about one-fifth of participants. Researchers did not include objective sleep measurements, such as those from sleep labs, relying instead on clinical interviews and self-reports. Prior cannabis use among participants was not systematically tracked, and because THC’s psychoactive effects make blinding difficult, some patients likely guessed their treatment assignment, which may have colored self-reported results. The findings, drawn mostly from female civilian survivors of physical or sexual assault, may not generalize to combat veterans or other trauma populations.
Funding and Disclosures
The trial was funded by an unrestricted grant from Bionorica SE, a German pharmaceutical company that also supplied the study medication; the funder had no role in designing the study, collecting or analyzing data, or deciding to publish the results. Several authors disclosed financial relationships with pharmaceutical companies, including consulting fees and research grants, though the study’s authors reported these were unrelated to the conduct of this particular trial.
Publication Details
The study, titled “Dronabinol for nightmares in post-traumatic stress disorder: a randomized controlled trial,” was authored by Stefan Roepke and colleagues from Charité–Universitätsmedizin Berlin and several partner institutions in Germany. It was published online August 5, 2026, in the journal Nature Medicine. DOI: 10.1038/s41591-026-04546-9. The trial is registered at ClinicalTrials.gov under identifier NCT04448808.
