Details

  • Number of Studies: 6
  • Sample Size: 286
  • Last Search Date: November 25th, 2025

Documentation

Data


Plain-language summary

This page summarizes what randomized clinical trials currently suggest about MDMA-assisted therapy for PTSD in simple terms.

  • What we asked: Do people with PTSD improve more after MDMA-assisted therapy than after a comparison condition (like a placebo or low dose), when both are paired with psychological support?
  • Who was studied: Adults with elevated PTSD symptoms. All trials were conducted by the Multidisciplinary Association for Psychedelic Studies (MAPS) and its spin-off Resilient Pharmaceuticals (formerly Lykos Therapeutics).
  • What was the intervention?: Trials gave MDMA at doses intended to produce a noticeable subjective experience, alongside structured psychotherapy before, during, and after dosing sessions. Participants received between two and three dosing sessions.
  • What we found: Across 6 randomized studies including 286 participants, MDMA was associated with larger improvements in PTSD scores than comparison conditions. People receiving MDMA were also more likely to meet definitions of treatment response and remission. More dosing sessions and a higher cumulative dose were associated with larger effects.
  • How confident are we?: We assigned an overall low certainty rating using the GRADE approach. Functional unblinding (participants and therapists guessing group assignment due to MDMA’s psychoactive effects) and the high rates of prior MDMA use among participants are key sources of concern. The true treatment effect in routine clinical practice may be smaller than what has been observed in trials so far.
  • What this does not mean: This is not medical advice. MDMA is not approved to treat PTSD. The FDA denied a new drug application for MDMA-assisted therapy in 2024, citing concerns about safety assessment, durability of effects, and potential expectancy bias. Most trials were small, carefully selected populations and may not generalize to broader treatment-seeking populations.
  • Safety note: This page focuses on whether PTSD symptoms improved. Adverse effects and safety were not the primary focus here and will be considered in a future analysis.
  • We will keep this updated: This is a living review. As new randomized controlled trials are published, we update our analyses and figures. For methods and full details, see the reproducibility guide and search strategy. An interactive overview is available on our dashboard.

About this page

Welcome to the landing page for our living systematic review and meta-analysis on MDMA-assisted therapy for PTSD (pre-registered here)! Follow the links in our documentation section for relevant information including our code-walkthrough (reproducibility guide). Here we provide an overview on our latest results, currently up-to-date with our most recent preprint.

Eligibility Criteria

We included randomized controlled clinical trials published in English in peer-reviewed journals comparing MDMA for PTSD with a comparator in adult (>18 years old) populations. In order to meet these criteria, the study population needed to include individuals with elevated PTSD symptoms; studies with only healthy participants were not considered. Eligible interventions included any dose and formulation of MDMA or other prodrugs of MDMA intended to produce an alteration of subjective experience in the patient, with or without the conjunctive use of therapy (e.g., microdosing studies were not included). Eligible comparators included any form of placebo with or without the conjunctive use of therapy, including lower doses of the intervention drug and any dose of other psychotropics intended to improve blinding (without known therapeutic efficacy for PTSD). Eligible comparators also included control for spontaneous improvement via waitlist or usual care. Additionally, for studies that used a crossover design, data from pre-crossover timepoints was required to account for potential carry-over effects of an MDMA dose.

Study selection

We identified 1,586 reports from our searches. Of these, 26 passed initial title and abstract screening and were reviewed as full-text articles. Six studies (286 participants) met inclusion criteria for our primary model.

Figure 1: PRISMA flow diagram showing study selection process

Study Characteristics

Our database consists of 61 effect sizes generated from 6 included studies, covering continuous and dichotomous measures of PTSD symptoms across timepoints (from four days after the first dose out to 12 months post-baseline). All studies included a psychotherapy or psychological support component before, during, and after MDMA dosing. See Table 1 for detailed study characteristics.

Table 1: Summary of RCTs on MDMA-assisted therapy for PTSD. N given is the number of participants randomized. Study endpoints are the reported primary endpoints for each study. The MDMA-AT protocol typically involves preparatory psychotherapy sessions, 2–3 dosing sessions lasting 8–10 hours spaced approximately one month apart, and integration psychotherapy sessions following each dosing session.

Risk of Bias

We assessed study bias using Cochrane’s Risk of Bias 2.0 tool, which is the standard approach for bias assessment in randomized controlled trials. We focused our assessment specifically on primary outcome variables. If information pertinent to assessment of risk of bias was not available, authors were contacted via email.

The RoB 2.0 tool examines five potential bias sources: randomization procedures, deviations from the intended protocol, missing outcome data, outcome measurement, and selective reporting. Each domain contains several questions rated on a four-level scale ranging from ‘yes’ to ‘no’, in addition to an option for insufficient information. Domain-level and overall study bias ratings are classified as low, medium, or high according to a predetermined algorithm. While this algorithm guides assessment, evaluators may justifiably override automated bias determinations when specific concerns warrant greater or lesser emphasis than the algorithm suggests.

Table 2: Additional study characteristics and risk of bias assessments using Cochrane’s Risk of Bias 2.0 tool domains. Randomization = bias due to randomization process; Deviations = bias due to deviations from intended interventions; Missingness = bias due to missing outcome data; Measurement = bias due to measurement of the outcome; Selection = bias due to selection of the reported results; Overall = overall risk of bias.

Overall, one study had some concerns (Mithoefer 2011) and five studies were deemed to have an overall low risk of bias (Table 2). Mithoefer 2011 had some concerns due to what appears to be a per-protocol study design, where dropouts were replaced with new enrollees. According to Cochrane guidelines, per-protocol study designs are not an appropriate analysis to estimate the effect of assignment to intervention. Although none of the studies had a high risk of bias, these assessments do not fully account for potential bias from functional unblinding or expectancy effects.

Results

MDMA treatment significantly reduces PTSD symptoms compared with control conditions

The analysis on continuous outcomes for the six studies included in the primary model showed a statistically significant reduction in PTSD scores after MDMA treatment compared with control conditions (Figure 2; Hedges’ g = -0.71 [-0.95; -0.47], p < 0.001, k = 6, n = 242), with low between-study heterogeneity (tau2 = 0.00 [0.00; 0.38], I2 = 0.0% [0.0%; 74.6%]).

Figure 2: Meta-analysis on continuous outcome variables. Boxes represent the standardized mean difference (Hedges’ g) for each study, and the lines extending from the box represent the 95% confidence interval around each effect size, while the size of each box is proportional to its weight. The diamond at the bottom represents the pooled effect size (meta-analytic mean). The gray line at the bottom represents the prediction interval of the expected range of true effects in a new study. HK = Knapp-Hartung adjustment.

Visual inspection of a funnel plot (SI Figure 1) revealed limited asymmetry. An Egger’s test did not find small study effects (intercept = -0.52 [-1.94; 0.91], t = -0.71, p = 0.52), although this test is underpowered given the small number of studies included in our meta-analysis (k < 10).

SI Figure 1: Funnel plot of 6 studies in primary meta-analytic model.

MDMA’s effects increase over dosing sessions and cumulative dose

Our three-level CHE model revealed a significant decrease in PTSD scores with MDMA compared to the control conditions consistent with our primary model (Hedges’ g = -0.60 [-0.86; -0.34], p < 0.001, k = 6, n = 286, tau2 = 0.00 [0.00; 0.23], I2 = 0.0% [0.0%; 66.0%]) (Figure 3a).

We next performed two separate meta-regressions. Adding the number of dosing sessions as a continuous predictor to our model (Figure 3b) revealed a significant effect of increased dosing sessions, such that more dosing sessions resulted in a greater between-group difference in PTSD scores (β = -0.20 Hedges’ g [-0.34; -0.06], p = 0.01). We also evaluated the cumulative MDMA dose as a continuous predictor; this revealed a significant effect of increased MDMA dose, such that a higher dose exposure resulted in a greater between-group difference in PTSD scores (β = -0.002 Hedges’ g/mg [-0.003; -0.0005], p = 0.01).

Figure 3: Three-level CHE model results. (a) Forest plot of the overall three-level model. (b) Meta-regression of the number of dosing sessions on effect size.

Higher response and remission rates, nonsignificant effect on depression symptoms after MDMA treatment

We found evidence for statistically significant greater treatment response with MDMA compared with control conditions (Figure 4a; RR = 1.35 [1.10; 1.66], p = 0.016, k = 5, n = 222), with low between-study heterogeneity (tau2 = 0.00 [0.00; 1.68]; I2 = 0.0% [0.0%; 79.2%]).

Figure 4a: Meta-analysis on response to treatment. Boxes represent the risk ratio (RR) for each study, and the lines extending from the box represent the 95% confidence interval around each effect size, while the size of each box is proportional to its weight. The diamond at the bottom represents the pooled effect size (meta-analytic mean). The gray line at the bottom represents the prediction interval of the expected range of true effects in a new study.

We also found that there were significantly higher remission rates with MDMA compared with control conditions (Figure 4b; RR = 2.25 [1.04; 4.87], p = 0.044, k = 4, n = 210), with low between-study heterogeneity (tau2 = 0.00 [0.00; 5.78]; I2 = 0.0% [0.0%; 84.7%]).

Figure 4b: Meta-analysis on remission rates. Boxes represent the risk ratio (RR) for each study, and the lines extending from the box represent the 95% confidence interval around each effect size, while the size of each box is proportional to its weight. The diamond at the bottom represents the pooled effect size (meta-analytic mean). The gray line at the bottom represents the prediction interval of the expected range of true effects in a new study.

In contrast, we did not find evidence for a significant effect of MDMA on co-morbid depression symptoms (Figure 4c; Hedges’ g = -0.66 [-2.70; 1.39], p = 0.30), with a low number of studies (k = 3, n = 118) and high heterogeneity (tau2 = 0.44 [0.00; 29.60], I2 = 68.0% [0.0%; 90.7%]).

Figure 4c: Meta-analysis on co-morbid depression symptoms in PTSD.

Sensitivity analyses

To assess the robustness of our primary findings, we performed the following sensitivity analyses:

  1. Alternate dosing in three-arm trials: Given that Mithoefer 2018 and Ot’alora G 2018 employed a three-arm design comparing MDMA at high and medium doses against a low-dose control, we conducted a sensitivity analysis substituting the medium dose intervention arm for the high-dose intervention arm used in our primary analysis.
  2. Fixed effect models: We ran fixed/common-effect models as sensitivity analyses to compare to random effects models. The fixed effect model assumes that the between-study variance (tau2) is 0, such that all studies share a common true effect size. For our continuous model, we used a standard inverse-variance weighting fixed-effect model on standardized mean differences (Hedges’ g). For our dichotomous models, we used a standard inverse-variance weighting fixed-effect model on the log risk ratio.
  3. Three-level model within-study correlation coefficient sweep: Our three-level CHE model assumed a constant within-study correlation coefficient (ρ) of 0.6. To test the sensitivity of our results against this approximation, we recalculated Hedges’ g as a function of ρ from 0 to 1 in 0.1 increments.
  4. Bayesian meta-analysis: We replicated our primary meta-analysis on continuous outcomes using a Bayesian implementation. We used “weakly informative” prior distributions for both the main effect and the heterogeneity parameter tau that have been recommended by prior work. The main effect prior was a normal distribution centered around 0, with a standard deviation of 1, while the tau prior was a half-normal distribution with a standard deviation of 0.5.

Sensitivity analyses provide convergent results

We performed a series of sensitivity analyses that supported our primary results. First, our model using medium-dose arms in place of the high-dose arms for three-arm trials showed a significant and comparable effect size (SI Figure 2; Hedges’ g = -0.75 [-1.19; -0.30], p = 0.008, k = 6, n = 234, tau2 = 5.31 x 10-6 [0.00; 3.11], I2 = 37.2% [0.0%; 75.0%]).

SI Figure 2: Forest plot of 6 studies using medium-dose arms in three-arm trials. Boxes represent the standardized mean difference (Hedges’ g) for each study, and the lines extending from the box represent the 95% confidence interval around each effect size, while the size of each box is proportional to its weight. The diamond at the bottom represents the pooled effect size (meta-analytic mean). The gray line at the bottom represents the prediction interval of the expected range of true effects in a new study. HK = Knapp-Hartung adjustment.

Furthermore, we replicated our primary analyses using a fixed effect model on continuous outcomes (Hedges’ g = -0.71 [-0.98; -0.45], p < 0.001, k = 6, n = 242), response outcomes (RR = 1.44 [1.20; 1.74], p < 0.001, k = 5, n = 222), and remission outcomes (RR = 2.49 [1.52; 4.09], p < 0.001, k = 4, n = 210).

Our Bayesian analysis revealed a Hedges’ g posterior distribution centered at -0.70 [-1.05; -0.36] (SI Figure 3).

SI Figure 3: Prior (dashed) and posterior (solid) distributions for (a) the pooled effect size (Hedges’ *g) and (b) the heterogeneity estimate (tau). Dark shading represents the 95% probability distribution.*

Finally, our three-level CHE results were consistent across a range of within-study correlation coefficients (SI Figure 4).

SI Figure 4: Hedges’ *g as a function of rho (ρ), the within-study correlation coefficient used in the three-level CHE model.*

GRADE Certainty of Evidence

Evidence derived from randomized controlled trials begins with a high certainty rating that can then be downgraded to moderate, low, or very low depending on assessments across four domains: risk of bias, inconsistency, indirectness, and imprecision. We downgraded the certainty of evidence by two levels for indirectness. Functional unblinding, particularly in the two large phase 3 trials which used an inert placebo as comparator, presents risks for expectancy effects to drive larger between-group differences in outcomes. Further, the percentage of participants with prior MDMA use in the studies was high (mean: 39%). Taken together, these factors of indirectness may lead to smaller treatment effects in a general population under routine clinical practice and contribute to an overall GRADE rating of low certainty. No downgrades were given for risk of bias, inconsistency, or imprecision.

Conclusions

Together, synthesis of studies to date suggest that MDMA-assisted psychotherapy results in reductions in PTSD symptoms and increased response and remission in patients with PTSD. However, our GRADE rating for the certainty of this evidence is low. Additional, large controlled trials with rigorous methods are needed, as are studies examining varying study characteristics (e.g., dosing sessions, longer follow-up durations) and expanding to more representative populations. As more RCTs are published, we will regularly update our SYPRES website and dashboard in a reproducible and transparent manner. This living systematic review, in conjunction with the associated open-science resources, aims to provide a valuable and transparent resource for researchers, clinicians, policymakers, and the public.

References

  1. Mithoefer, M.C., Wagner, M.T., Mithoefer, A.T., Jerome, L., Doblin, R. (2011). The safety and efficacy of {+/-}3,4-methylenedioxymethamphetamine-assisted psychotherapy in subjects with chronic, treatment-resistant posttraumatic stress disorder: the first randomized controlled pilot study. J Psychopharmacol, 25(4), 439-52. https://doi.org/10.1177/0269881110378371
  2. Oehen, P., Traber, R., Widmer, V., Schnyder, U. (2013). A randomized, controlled pilot study of MDMA (± 3,4-Methylenedioxymethamphetamine)-assisted psychotherapy for treatment of resistant, chronic Post-Traumatic Stress Disorder (PTSD). J Psychopharmacol, 27(1), 40-52. https://doi.org/10.1177/0269881112464827
  3. Mithoefer, M.C., Mithoefer, A.T., Feduccia, A.A., Jerome, L., Wagner, M., Wymer, J., Holland, J., Hamilton, S., Yazar-Klosinski, B., Emerson, A., Doblin, R. (2018). 3,4-methylenedioxymethamphetamine (MDMA)-assisted psychotherapy for post-traumatic stress disorder in military veterans, firefighters, and police officers: a randomised, double-blind, dose-response, phase 2 clinical trial. Lancet Psychiatry, 5(6), 486-497. https://doi.org/10.1016/S2215-0366(18)30135-4
  4. Ot'alora G, M., Grigsby, J., Poulter, B., Van DerVeer, J.W. 3rd., Giron, S.G., Jerome, L., Feduccia, A.A., Hamilton, S., Yazar-Klosinski, B., Emerson, A., Mithoefer, M.C., Doblin, R. (2018). 3,4-Methylenedioxymethamphetamine-assisted psychotherapy for treatment of chronic posttraumatic stress disorder: A randomized phase 2 controlled trial. J Psychopharmacol, 32(12), 1295-1307. https://doi.org/10.1177/0269881118806297
  5. Mitchell, J.M., Bogenschutz, M., Lilienstein, A., Harrison, C., Kleiman, S., Parker-Guilbert, K., Ot'alora G, M., Garas, W., Paleos, C., Gorman, I., Nicholas, C., Mithoefer, M., Carlin, S., Poulter, B., Mithoefer, A., Quevedo, S., Wells, G., Klaire, S.S., van der Kolk, B., Tzarfaty, K., Amiaz, R., Worthy, R., Shannon, S., Woolley, J.D., Marta, C., Gelfand, Y., Hapke, E., Amar, S., Wallach, Y., Brown, R., Hamilton, S., Wang, J.B., Coker, A., Matthews, R., de Boer, A., Yazar-Klosinski, B., Emerson, A., Doblin, R. (2021). MDMA-assisted therapy for severe PTSD: a randomized, double-blind, placebo-controlled phase 3 study. Nat Med, 27(6), 1025-1033. https://doi.org/10.1038/s41591-021-01336-3
  6. Mitchell, J.M., Ot'alora G, M., van der Kolk, B., Shannon, S., Bogenschutz, M., Gelfand, Y., Paleos, C., Nicholas, C.R., Quevedo, S., Balliett, B., Hamilton, S., Mithoefer, M., Kleiman, S., Parker-Guilbert, K., Tzarfaty, K., Harrison, C., de Boer, A., Doblin, R., Yazar-Klosinski, B. (2023). MDMA-assisted therapy for moderate to severe PTSD: a randomized, placebo-controlled phase 3 trial. Nat Med, 29(10), 2473-2480, https://doi.org/10.1038/s41591-023-02565-4