A nightly dose of prescription THC reduced trauma nightmares more than a placebo did in a 171-patient randomized trial led by Charité in Berlin, published August 5, 2026 in Nature Medicine. On the trial’s clinician-rated nightmare measure, scores fell further with the drug than with placebo over ten weeks.
Key takeaways
- Clinician-rated nightmare scores fell 1.50 points further with dronabinol than with placebo over ten weeks.
- What was measured was nightmares specifically, not overall PTSD severity.
- Serious side effects occurred in 7 of 87 patients on the drug (8.0 percent) and in none on placebo.
What the study found
Researchers at Charité, working with centers in Hamburg and Mannheim, randomly assigned 171 adults who had PTSD and recurrent nightmares to either dronabinol, a plant-derived THC, or a matching placebo. Both were taken once daily before bedtime for ten weeks, at doses of 2.5 to 15 mg. Neither patients nor clinicians knew who got what. Participants averaged 37.9 years old and 79.3 percent were women, with 87 assigned to dronabinol and 84 to placebo.
The main measure was the CAPS-IV B2 item, a clinician-scored rating of how often nightmares happen and how intense they are. According to the study, scores dropped 1.50 points further with dronabinol than with placebo. The true difference is very likely between 0.71 and 2.28 points in the drug’s favor (95 percent confidence interval), and there is less than a 1 in 1,000 chance the result is coincidence (P < 0.001). In statistical terms the effect was moderate in size (Cohen’s d = 0.65).
Side effects were common in both arms: 89.7 percent of the dronabinol group and 77.1 percent of the placebo group, a gap of about 13 percentage points. Serious adverse events occurred in seven patients on dronabinol (8.0 percent) and none on placebo. Slightly fewer people stopped the drug because of side effects (5.7 percent) than stopped placebo (7.2 percent).
Dr. Kumar’s take
Here is what makes this different from most promising trial news. Dronabinol is not a future drug. It is a prescription cannabinoid already on the market, and a US physician can write for it off-label starting tomorrow. That is exactly why I want to be precise about what this trial did and did not show.
It showed a nightmare effect, not a PTSD cure. The one thing this trial was built to answer was whether a nightly cannabinoid reduces the frequency and intensity of trauma nightmares, and that is the single item it measured as its primary endpoint. Overall PTSD severity was not what the headline result describes. Coverage is already drifting toward “cannabis treats PTSD.” That is not what was tested and not what the paper claims.
The placebo arm matters too. This was a double-blind, placebo-controlled trial, so the 1.50-point gap is what is left after attention, expectation, and time have done their work. I have written before about how placebos helped older adults even when they knew the pill was fake, and sleep symptoms respond strongly to that effect.
Then the boundaries. Ten weeks is short for a drug someone might take for years, and the authors say plainly that long-term efficacy and safety still require further evaluation. The cohort averaged 37.9 years and was nearly 80 percent women, which does not look like the older, mostly male veteran population where trauma nightmares get treated most often in the US. And seven serious adverse events in the drug arm against zero on placebo is a real signal in a trial this size. Attrition is worth knowing about too: of the 171 people randomized, the week-10 nightmare score was actually available for 76 of the 87 on dronabinol and 69 of the 84 on placebo, so 26 patients were missing from the main comparison. The work was supported by Bionorica SE.
What it means for you
If you have PTSD with nightmares that have not responded to standard treatment, this is worth raising with your psychiatrist as a real option rather than a rumor. Ask about the nightmare target specifically, not PTSD in general, and about the drug’s controlled-substance status and how coming off it would work.
What this trial does not support is self-medicating with cannabis. The study used a measured prescription dose taken once before bed, titrated by clinicians, in a monitored setting. Smoked or edible cannabis is a different exposure with a different dose curve and a different risk profile.
Trauma-focused psychotherapy remains the treatment with the strongest evidence for PTSD as a whole. A drug that gives you back your nights is a meaningful addition, not a replacement. If your problem is broader sleep quality rather than trauma nightmares, I looked at a different option in saffron extract improves sleep quality: randomized double-blind trial.