A deep dive into the magnesium-ibogaine treatment.

So What Does a 12-Month Ibogaine Study Actually Show?

A newly peer-reviewed Stanford study asks a simple question: “Is ibogaine treatment durable?” On the surface, its answer is remarkable.

Thirty male US Special Operations veterans with histories of traumatic brain injury had undergone a magnesium–ibogaine treatment programme in Mexico. At twelve months, 25 completed the principal follow-up assessment. Disability, PTSD, depression and anxiety scores remained dramatically below baseline, with effect sizes of Cohen’s d ≥ 2.18. Among participants who had entered remission immediately after treatment, the estimated probability of remaining in remission at twelve months was 84% for PTSD, 66% for depression and 61% for anxiety.

The journal paper appeared in Translational Psychiatry on 12 August 2026, although the findings themselves are not entirely new: a preprint was posted to Research Square on 17 December 2025.

Those are impressive results. But there is another number in the paper that changes what they can reasonably be taken to mean.

Of the 25 participants assessed at twelve months, 23 had used at least one other psychedelic after the one-month follow-up. Nineteen had used 5-MeO-DMT, 18 of them in a retreat setting. Twelve had undertaken therapy or counselling. Only three reported psychotropic medication use.

That does not make the study worthless. It does, however, change the question it is capable of answering.

The most interesting part of the design happened during the first month

The original MISTIC study, published in Nature Medicine in 2024, followed the same 30 veterans. Twenty-eight had histories classified as mild TBI; only two reported moderate or moderately severe injuries. At baseline, 23 met criteria for PTSD, 15 for major depressive disorder and 14 for an anxiety disorder. The average participant reported an extraordinary 38.6 previous TBIs, although that figure had enormous variation, and the most severe injury had occurred on average more than fifteen years earlier.
These were not veterans randomly recruited by Stanford and assigned to ibogaine. They had independently elected to undergo treatment at Ambio Life Sciences in Mexico, facilitated by the nonprofit Veterans Exploring Treatment Solutions, or VETS. VETS informed potential participants about the research and referred interested individuals to Stanford. VETS also funded their treatment, travel and accommodation. Stanford conducted the psychiatric and neuroimaging assessments, while Ambio conducted the treatment.

That creates obvious selection and expectancy problems. These were people who had already decided that an unconventional psychedelic intervention in Mexico was worth pursuing.

But the design also contains a feature that deserves more attention than it usually receives.

Ambio's programme ordinarily follows ibogaine several days later with 5-MeO-DMT. For the Stanford study, participants specifically refrained from taking 5-MeO-DMT until after the one-month assessments had been completed.

That is important.

It means that although the original study was still open-label, uncontrolled and embedded in a complicated treatment environment, the researchers deliberately created a relatively cleaner first-month window. For that period, at least one major additional psychedelic normally included in the programme was intentionally removed.

And during that month the changes were enormous.

At one month, Cohen’s d was 2.20 for disability, 2.54 for PTSD, 2.80 for depression and 2.13 for anxiety.

That is arguably the most interesting signal in the entire MISTIC project.

The twelve-month follow-up then does something fundamentally different. It stops observing participants inside that deliberately constrained period and follows them back into their actual lives.

That is not researcher sloppiness. It is an unavoidable feature of naturalistic long-term follow-up.

The methodological problem arises if persistence of improvement in the people becomes interpreted as persistence of a specific pharmacological effect of ibogaine.

Those are not the same proposition.

During months one to twelve, 92% of the participants available at the final follow-up used another psychedelic. Most used 5-MeO-DMT. Many returned to retreat settings. Nearly half received therapy or counselling. The paper additionally lists psilocybin, ayahuasca and even ibogaine itself among the post-treatment interventions examined, although the main article does not state how many participants took ibogaine again.

So the twelve-month study does not observe a clean sequence of:

ibogaine → nothing else → twelve-month outcome.

It observes people who received ibogaine, improved dramatically during the first month, and then spent the following eleven months living complicated lives in which almost all of them took further psychedelics and many pursued additional therapeutic interventions.

That still tells us something. It tells us their improvement did not simply vanish.

It cannot tell us how much of that persistence belongs specifically to the original ibogaine treatment.

The subgroup analysis does not solve this

The authors address the problem rather than hiding it. They report that long-term outcomes appeared similar among participants who did and did not pursue additional interventions and describe exploratory subgroup analyses stratified by particular intervention types.

That is worth reporting fairly.

But it is easy to give this analysis more weight than the sample can carry.

For psychotherapy or counselling, the split is reasonably intelligible: 12 participants received it and 13 did not.

For 5-MeO-DMT, however, the comparison is 19 versus 6.

And for the broadest fact in the paper — whether somebody took any additional psychedelic — the descriptive split is 23 versus 2.

The authors do not claim in the main paper to have conducted a definitive 23-versus-2 causal test, and it would be wrong to suggest otherwise. Their analyses concern specific intervention categories. But the fact remains that only two of the 25 twelve-month completers constitute an entirely psychedelic-free comparison group.

That makes the broad claim that subsequent psychedelic exposure did not explain the durability essentially impossible to establish from this dataset.

The authors themselves call the psychedelic use “an important limitation” and state that they did not control for specific treatments, activities or significant life events during the follow-up period. They explicitly say that future studies are needed to identify which factors interacted with the apparent durability.

That acknowledgment matters, because this is not really a dispute with the researchers over what their design can prove. Their own abstract calls the findings preliminary and says randomized controlled trials are required to clarify the causal contributions of ibogaine, subsequent interventions and their combination.

The danger comes when the spectacular numbers are detached from that sentence.

A Cohen's d of 3.35 does not mean ibogaine caused a 3.35 effect

The statistics themselves deserve similar care.

At twelve months, Cohen’s d was 2.27 for disability, 2.72 for PTSD, 3.35 for depression and 2.18 for anxiety. Every corrected p value was below .001.

These are enormous effect sizes.

But the paper tells us precisely what they represent: the contrast between each follow-up time point and baseline in the same observational cohort.

They are not ibogaine-versus-placebo effect sizes.

That distinction is critical.

A pre-post effect size measures how far the observed scores moved. In an uncontrolled study, it cannot tell us how much of that movement was caused by the treatment. The number contains everything that changed between the two measurements: the ibogaine experience, the retreat and coaching environment, expectancy, spontaneous symptom fluctuation, regression towards the mean, later psychedelics, psychotherapy, medications and ordinary life.

An enormous d therefore makes one conclusion very convincing:

These people's measured symptoms changed enormously.

It does not automatically make another conclusion convincing:

Ibogaine caused that entire change.

No amount of statistical significance can create the counterfactual population the study never observed.

The missing data are strange rather than simply “five dropouts”

It is also slightly misleading to describe the study as simply losing five of its thirty participants.

Twenty-five did complete the twelve-month assessment, but the intermediate follow-ups were much patchier. At three months only ten participants supplied the psychiatric assessments and eight supplied the disability measure. At six months the figures were 21 and 17; at nine months, 17 and 15. Participants therefore disappeared and later returned rather than simply dropping out permanently.

The researchers used linear mixed-effects models, which are designed to make use of incomplete repeated-measures datasets, and 27 of the original 30 supplied at least one longer-term assessment.

That is statistically reasonable.

It still does not make missingness irrelevant. Mixed models can use the observations that exist; they cannot reveal what unobserved measurements would have shown or guarantee that the reasons people missed assessments were unrelated to their outcomes.

With a starting sample of thirty, individual participants matter.

One finding deserves to be stated clearly in ibogaine's favour: acute safety

A critical appraisal should not only count evidence in one direction.

Ibogaine has a genuine and well-documented cardiac risk, particularly QT prolongation and potentially fatal arrhythmia. The original MISTIC study therefore used medical screening, magnesium administration, continuous cardiac monitoring and intensive medical supervision.

Within this highly selected and carefully monitored group, there were no serious or unexpected treatment-emergent adverse events and no clinically meaningful QT prolongation, bradycardia, tachycardia or haemodynamic instability. All participants experienced temporary cerebellar effects such as mild ataxia or intention tremor that resolved within 24 hours; headache, nausea and smaller numbers of anxiety, hypertension and insomnia were also treated.

That is genuinely useful evidence.

It does not establish that ibogaine is generally safe, particularly because people with cardiovascular, liver and kidney problems were excluded and treatment occurred under unusually intensive monitoring. But it provides meaningful preliminary evidence that this specific magnesium–ibogaine protocol can be delivered without serious acute events in a carefully screened Special Operations veteran population.

That finding may ultimately prove more robust than some of the efficacy claims.

What about claims that their brains were healing?

The twelve-month paper should also not be criticised for something it does not actually claim to demonstrate.

Its principal outcomes are disability and psychiatric symptoms. It does not present twelve-month imaging evidence showing that damaged brains were biologically repaired.

There are, however, separate publications from the same cohort that make the neuroscience relevant.

A 2026 functional-neuroimaging study reported changes in regional cerebral blood flow and functional connectivity through one month, including associations between changes in the left insula and anterior cingulate and improvements in disability. Importantly, those authors describe the findings as hypothesis-generating and say larger controlled neuroimaging studies are required to validate them.

Another 2026 paper analysed veterans' written descriptions of the experience and identified what it called an “embodied sense of neural repair.” That phrase describes what participants felt. The authors themselves explicitly warn that these experiences should primarily be understood as embodied metaphors and that the convergence with cognitive and imaging changes does not mean participants were literally feeling neurobiological repair taking place.

That distinction is worth keeping.

There are intriguing neurological findings around this cohort. There is not yet controlled evidence that ibogaine reverses traumatic brain injury.

Who funded it, and what interests are disclosed?

There is no need to imply a financial scandal where the evidence does not support one.

VETS funded participants' ibogaine treatment, travel and accommodation. Stanford states that it received no funding from VETS or Ambio for the original study; that component was supported independently by private philanthropy.

The twelve-month paper acknowledges funding from Steve and Genevieve Jurvetson, the Effie and Wofford Cain Foundation, Eugene Jhong, the Saisei Foundation, Laura Keller, collaboration with VETS and a Canadian Institutes of Health Research fellowship. The authors state that funders had no role in study design, execution, analysis or manuscript preparation.

There are nevertheless relevant commercial interests.

Stanford owns intellectual property relating to magnesium–ibogaine. Nolan Williams, Ian Kratter, Andrew Geoly and John Coetzee are named inventors on Stanford-owned magnesium–ibogaine intellectual property. The paper also discloses Williams's advisory and equity interests and states that Kratter receives a salary from Soneira and owns Soneira equity or stock options.

Those disclosures are not evidence that results were manipulated. They are precisely why conflict-of-interest declarations exist.

They do mean that independent replication matters.

The twelve-month paper answers a narrower question than its title suggests

There is a genuinely striking result here, and dismissing it would be as unhelpful as overselling it.

Thirty highly selected Special Operations veterans entered the programme with longstanding psychiatric and functional difficulties. Their measured symptoms fell dramatically after treatment. The one-month results occurred during a deliberately constrained period in which participants had specifically refrained from taking the 5-MeO-DMT that ordinarily followed ibogaine at the clinic. No serious acute adverse events occurred under the study's screening and monitoring protocol. Those are legitimate signals deserving serious controlled research.

A year later, most of the veterans who were assessed were still doing dramatically better.

That is what the new study establishes.

But between the relatively clean one-month window and the twelve-month endpoint, 23 of 25 participants took another psychedelic. Nineteen used 5-MeO-DMT. Eighteen did so in retreat settings. Twelve undertook therapy or counselling. Ibogaine itself appears among the additional interventions recorded. Five of the original thirty were absent from the final assessment, and the intermediate follow-ups contained substantial missing data. There was still no untreated, placebo or treatment-as-usual comparison group.

The interesting scientific story is therefore not that the study has shown nothing.

It is almost the opposite.

The original one-month result may tell us considerably more about an ibogaine-containing intervention than the spectacular twelve-month numbers do.

Once these veterans returned to ordinary life and almost universally continued experimenting with psychedelics and other forms of treatment, the follow-up became evidence about the durability of their improvement, not clean evidence about the durability of ibogaine's specific effect.

The authors largely acknowledge that distinction themselves.

And perhaps the single most important number for anyone repeating the claim that one ibogaine treatment produced twelve months of benefit is not d = 3.35, p < .001 or 84% remission.

It is 23 out of 25.

Sources

Faerman, A., Lissemore, J.I., Geoly, A.D. et al. (2026). Is ibogaine treatment durable? 12-month follow-up of magnesium–ibogaine therapy (MISTIC) in special operations veterans with traumatic brain injuries. Translational Psychiatry. DOI: https://doi.org/10.1038/s41398-026-04327-5

Cherian, K.N., Keynan, J.N., Anker, L. et al. (2024). Magnesium–ibogaine therapy in veterans with traumatic brain injuries. Nature Medicine, 30, 373–381. DOI: https://doi.org/10.1038/s41591-023-02705-w

Rolle, C., Williams, N., Faerman, A. et al. (2025). Is ibogaine treatment durable? Research Square preprint, posted 17 December 2025. DOI: https://doi.org/10.21203/rs.3.rs-6909189/v1

Sridhar, M. et al. (2026). Neural Correlates of Ibogaine: Evidence From Functional Neuroimaging of Military Veterans. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 11(6), 624–633. DOI: https://doi.org/10.1016/j.bpsc.2026.02.001

Olash, C., Buchanan, D.M., Brown, R. et al. (2026). Accelerated recovery using magnesium ibogaine: characterizing the subjective experience of its rapid healing from neuropsychiatric disorders. npj Mental Health Research, 5, 8. DOI: https://doi.org/10.1038/s44184-026-00185-7

#Ibogaine #5MeODMT #PsychedelicResearch #TraumaticBrainInjury #PsychedelicScience