The Inner Cosmos · The Doors of Perception

Ibogaine: The Root That Interrupts Addiction

A shrub from the forests of Central West Africa carries an alkaloid in its root bark that, in open-label case series, has been recorded ending the signs of opioid withdrawal within a day in 25 of 33 cases. The same alkaloid, and more so the metabolite the liver makes from it, blocks the potassium channel the heart uses to reset itself between beats, and a peer-reviewed forensic review counted nineteen deaths between 1990 and 2008. No randomised, placebo-controlled trial of its efficacy has ever been completed. Both halves of that are the subject here, and the studies most often cited for the first half each contain, in their own results, someone who did not survive.

CASE Y_1_10 Reliability: The plant, the pharmacology and the cardiac mechanism are established (Tier 1); every human efficacy result is uncontrolled 22 Sources

Tier 1 · Verified Tier 2 · Credible Tier 3 · Speculative Tier 4 · Dubious

Two things are true of the same molecule, and neither one cancels the other. In a series of 33 open-label cases published in 1999, the signs of opioid withdrawal resolved without further drug-seeking within 24 hours in 25 people, and held through a 72-hour observation. In a peer-reviewed forensic review covering 1990 through 2008, 19 people died within hours or days of taking that same compound. Both records are in the published literature. This article is about what kind of record each one is.

The order below is deliberate. What happens to withdrawal in the first days, what happens to abstinence over the following year, and what ibogaine does to the heart are three different questions resting on three different bodies of evidence, and folding them into one sentence would be three errors at once. They are kept apart throughout. The safety material is not a closing caveat here: it holds two full sections, a table of its own, and most of the article's figures.

One more thing about the plan, stated once and not repeated. No dose, preparation, route, schedule or screening procedure appears anywhere in this article, in any section, caption or note. That is an editorial decision rather than an omission. This is a piece about a research subject and about a living religion, and it is not usable as instructions.

Two neighbouring subjects are not this one. This wing's article on ego dissolution covers what the classical psychedelics do to the sense of self, and ibogaine is not a classical serotonin-2A drug, so that material stays there. The Healing Arts article on the psychedelic renaissance covers psilocybin, MDMA and the trials that carried them back toward medicine. Ibogaine sits outside that story on safety grounds, and section 08 is where the reason is set out.

01The Plant, And What Is In Its Root Bark

The botany is uncontested, so it is worth setting down first.

Two objects run through this article and they are not the same thing. One is prepared plant material, a mixture of dozens of alkaloids. The other is ibogaine, the single compound isolated from within it, which is what the pharmacology and the cardiac electrophysiology below are about. Keeping the two apart matters, and section 08 is where it matters most.

02Bwiti, And What Iboga Is For

Bwiti is not a prologue to a pharmacology story. It is a living religion with its own account of what iboga is for, and that account is older than every piece of research below it.

Reports describe the initiatory state lasting a day or more. This article gives no range for it, because our two research files give different durations at different tiers and averaging them would produce a figure neither one supports. Section 12 carries the detail.

Secondary accounts place that declaration in June of that year, in an act of the Council of Ministers. No primary government document confirming the day was located for this article, so only the year is stated above.

Tier 2 · Credible

The traditional Bwiti context provides psychological support and community integration that is absent from most Western clinical settings.

That runs in a direction worth noticing. The ceremonial container is something the clinic lacks, rather than folklore the clinic improved on.

A word about what is not pictured here, since silence about it would be its own kind of claim. No image of a Bwiti ceremony appears in this article. The openly licensed photographs located for this article show a building, museum objects, or an unnamed and identifiable individual whose photographer describes the religion in terms this article will not repeat and whose consent is not recorded anywhere on the file. This passage runs on the building and the object, and where that leaves a gap, the gap is left. A living religion does not need a face to be treated seriously.

03The Nineteen-Year-Old In New York

That sequence is the origin of the entire field and it should be read for exactly what it is: an anecdote, followed by a self-administered series of seven, run by an interested party. It is historically important and it is not evidence of efficacy. Everything in sections 05 and 06 exists because of it, and none of it inherits its authority.

04A Molecule That Does Many Things At Once

Ibogaine's pharmacology is unusual in a specific and consequential way, and that difference is what keeps this article off the territory of the classical psychedelics.

The plain consequence is that ibogaine is not a classical serotonin-2A psychedelic. Whatever it is doing, it is not doing it the way those drugs do, and results from that literature do not carry over here by default.

05What Happens To Withdrawal In The First Days

This is the first of the three questions and the one with the strongest observational record behind it. It concerns the acute interval, the first day to three days after ingestion. It is not the same question as whether anyone is still abstinent a year later, and the two are not answered by the same evidence.

That last outcome is not a footnote and it is not an outside critic's addition. It sits in the paper's own list of results, in the same abstract as the number the whole field quotes. Neither of our two research files mentions it.

06What Happens Over The Following Year

The second question is a different one and the evidence for it is thinner. Losing the signs of withdrawal within 24 hours and being off opioids twelve months later are separate outcomes, and the studies reporting the second are small, uncontrolled and heavy with attrition.

That is the second death, and it belongs in the same breath as the twelve-month result rather than in a safety section further down the page. The twelve-month finding comes from eight people, inside a cohort of fourteen, one of whom did not survive the treatment. Our primary research file describes this study as a prospective observational study in New Zealand showing significant reduction in opioid use at 12 months, and stops there.

07What Ibogaine Does To The Heart

This is the third question, and the point at which ibogaine separates from the other compounds in the psychedelic-therapy conversation. It is not a caveat attached to the two sections above. It is a body of mechanistic evidence at least as strong as anything in them, and in our secondary research file it holds its own numbered Tier-1 section, which the primary file does not give it.

The recovery phase is the thing this article keeps returning to. The heart's electrical cycle runs beat by beat, each beat followed by a recovery before the next can be organised, and the QT interval is the measure of how long that recovery takes. Everything ibogaine does to the heart, in this section and the next, is a change in that one span.

That is the fact in this article a reader is least likely to arrive already holding, and it is the shape of the whole trade-off. The molecule credited with the sustained anti-craving benefit is the same molecule carrying the sustained cardiac risk. The danger outlasts the experience.

Our two research files point in opposite directions at exactly this point, and the disagreement is about the safety-critical molecule. The primary file proposes noribogaine, at Tier 2, as a potentially safer alternative to ibogaine with a more favourable cardiac profile. The secondary file states at Tier 1 that noribogaine's greater hERG potency and longer half-life make it the primary cardiac concern. This article carries the Tier-1 position, which is also the more dangerous of the two readings, and does not carry the safer-alternative framing anywhere.

An independent scoping review published in Molecules in 2026, by Esperanca, Gomes and Campos, states the same limitation in more detail and is more current than either of our files. Ibogaine's principal safety limitation, in that review's account, is its propensity to induce cardiac electrical instability through concentration-dependent prolongation of ventricular repolarisation, by inhibition of the hERG channel, reducing repolarisation reserve and facilitating early afterdepolarisations. Clinical cases are reported with corrected QT intervals exceeding 600 milliseconds, one of them reaching 714. A concentration-dependent increase in corrected QT associated with noribogaine exposure has been established in controlled human studies. And the prolongation persists for several days after ingestion, despite the clearance of the parent compound.

That last clause is the one to carry out of this section. Whatever the visionary experience is or is not doing, it is over long before the cardiac risk is.

08The Fatality Record, And The Rate Nobody Can Give

The section above is about what can happen. This one is about what has been recorded as happening, and it is a different kind of evidence again: a forensic case series, real in every individual case and without a denominator underneath it.

The same review identified two further risk factors beyond pre-existing cardiovascular disease and co-used substances: seizures associated with withdrawal from alcohol and benzodiazepines, and uninformed use of ethnopharmacological forms of iboga, meaning plant material of unknown alkaloid content rather than a measured preparation. That is an observation about the shape of the record, not a list of things to arrange differently. The deaths clustered around people whose bodies or whose circumstances were already compromised, and that is precisely the population this treatment attracts.

A rate is what a reader will want at this point, and this article does not have one to give. Our secondary file states that an estimated 1 in 300 ibogaine treatments results in death, and that there were approximately 30 documented fatalities by 2020. Neither figure could be traced to a peer-reviewed derivation in the checking done for this article, so neither is carried as fact. What the peer-reviewed forensic record documents is 19 deaths across the two decades to 2008. The 2026 scoping review declines to give an overall mortality rate at all, and says the evidence base is inherently incomplete because ibogaine use often occurs outside regulated medical settings. Alper, Lotsof and Kaplan estimated in 2008 that 3,414 individuals had taken ibogaine as of February 2006, which is the kind of denominator such ratios are usually built from, and that paper offers no mortality estimate of its own.

The refusal runs in both directions, and the second direction matters more than the first. Declining to state a rate is not a suggestion that the risk is small. The uncertainty here runs toward more risk rather than less, for the plain reason that most ibogaine is taken in places where a death would never enter a systematic record. The honest position is that the true number of deaths is unknown and probably undercounted.

09The Trial That Has Never Been Run

That is our own files stating their strongest limiting claim in their own voice, which means this article can state it at full strength too. It is also still true. The 2026 scoping review says explicitly that to date no randomised placebo-controlled clinical trials have evaluated ibogaine's efficacy relative to placebo, that all available human efficacy data derive exclusively from open-label observational studies, retrospective analyses and case reports, and that ongoing trials are registered but their results remain unavailable.

The clause worth holding onto inside the file's own sentence is the one granting that the efficacy claims are supported by consistent observational data. The honest verdict is not that the evidence is worthless. It is that it is the wrong kind of evidence to settle the question, and no quantity of it converts into the other kind.

10The Law, And The Sector It Produced

Those countries are named as a description of where the unregulated sector exists, and not as a list of destinations. No clinic is named anywhere in this article. The two facts that matter belong side by side: the literature says cardiac screening, monitoring and professional supervision are essential, and most ibogaine treatment happens in exactly the places least able to guarantee any of the three. The risk is highest where the oversight is weakest, and that is where the treatment mostly is.

New Zealand is in that list for a second reason as well. The twelve-month observational study in section 06 was conducted there because the legal status of ibogaine permitted it, and the paper's own framing says that legal availability enabled better collaboration with providers, and describes the resulting outcome evaluation as unique. That is the narrow and honest policy point available here: legality is what makes the research possible. It is not an argument about what any law should be.

Something changed recently that neither of our research files carries. Texas Senate Bill 2308, of the 89th Legislature's Regular Session, establishes a consortium to conduct United States Food and Drug Administration drug-development clinical trials with ibogaine, to seek approval of its use for opioid use disorder, for co-occurring substance use disorder, and for any other neurological or mental health conditions for which ibogaine demonstrates efficacy, and to govern the administration of that treatment. The Governor signed it on 11 June 2025 and it took effect immediately. A United States state has put public money behind pushing a Schedule I compound through the trials that have never been run. The primary bill history is linked below. A funding figure circulated in press coverage and is not stated here, because it does not appear on that page.

11The Compounds Built To Keep The Effect And Drop The Risk

That is our own primary file saying, in its own voice, that an entire drug-development programme rests on an unproven premise.

Our secondary file states that 18-MC entered Phase II clinical trials under the company MindMed by 2023. The direction of travel is backwards in that sentence. MindMed's own dated release of 19 May 2022 reports topline data from a Phase 1 placebo-controlled trial of MM-110, also called zolunicant or 18-MC, in 108 healthy volunteers, describing the compound as an alpha-3-beta-4 nicotinic cholinergic receptor antagonist and a non-hallucinogenic congener of ibogaine, and states that a Phase 2a trial was then intended. Trade reporting says that after a request from the Food and Drug Administration for additional preclinical characterisation the company discontinued active development in 2023 and moved its resources elsewhere. That last step is trade press rather than a primary filing, so it is carried here as reported rather than as established. The analog designed to keep the benefit and drop the heart risk got through a Phase 1 safety study and then stalled.

The most significant recent development that neither of our files carries is organised around the same problem. Cherian, Keynan, Anker and colleagues, with Nolan R. Williams as senior author, published a study in Nature Medicine in 2024 of magnesium-ibogaine therapy in veterans with traumatic brain injuries: a prospective observational study of 30 male Special Operations Forces veterans with predominantly mild traumatic brain injury, provided together with complementary treatment modalities. The stated rationale for the magnesium is that ibogaine has been associated with fatal cardiac arrhythmia, and that co-administering magnesium may mitigate that risk.

Two things about that study should be said plainly. It has no control group and the intervention was bundled with other treatments, which makes attribution to ibogaine alone impossible, and no outcome percentages from it are quoted in this article. And the magnesium is a proposal rather than an established mitigation: the paper's own word is may. What the study does show is where the frontier of this field now sits. The current research programme is organised around trying to blunt the cardiac risk, which is a fair measure of how large that risk is understood to be by the people closest to it.

That is a different claim from the depression finding in section 05, which was a Tier-1 observation of lower self-reported depressive symptoms inside a single open-label detoxification study, and not an evaluation of ibogaine as an antidepressant. The two are not merged here.

12Where Our Own Files Are Wrong

The standing rule in this wing is that our own sources get checked like anybody else's. Two research documents back this article, and both are recent, both score well on their own headers, and both carry errors a reader would have no way to see. The list is here rather than buried, and the corrected versions are what the article above uses.

The most serious is a citation. Our secondary file gives the ibogaine-fatalities review as a paper in Drug and Alcohol Dependence and attaches a digital object identifier to it. That identifier resolves to a paper about depressive mood and tobacco use, by different authors, on an unrelated subject. The journal named in the row is wrong as well. The real paper is Alper, Stajic and Gill in the Journal of Forensic Sciences, and it is the source of the nineteen-death figure this article uses. A wrong identifier on the single most safety-critical citation in a pair of files is the worst place in the pair for one to sit. The corrected identifier is in the sources below.

A second identifier fails the same way. Our primary file attaches a digital object identifier to James Fernandez's 1982 book on Bwiti, and it resolves to a 1986 review of that book in a journal, not to the book. The book is cited above by its international standard book number instead. The same file also misspells the author's surname.

Three of the efficacy attributions are wrong. Our secondary file credits the 33-case series, and the 25 people in it whose withdrawal signs resolved, to Deborah Mash's Saint Kitts work; they belong to Alper and colleagues in 1999. It describes those cases as having been conducted at a treatment clinic; the paper says non-medical settings, open label. And it places Brown and Alper's 30-subject study in New Zealand, which is a different study altogether, while moving that study's half-the-group result from one month to twelve. All three are corrected in sections 05 and 06.

Two studies contain a death that neither file mentions. Alper's 1999 series lists one fatality among its own outcomes. The New Zealand cohort of fourteen lost one participant during treatment. Both are stated above at the point where those studies are used, which is where they belong, and not gathered into a safety appendix where they would read as somebody else's objection rather than as part of the result.

Two numbers our files disagree about are not given here at all. They disagree on how long the Bwiti initiatory state lasts, at different tiers, so this article says a day or more and gives no range. They disagree on noribogaine's half-life, at the same tier, so this article says it lasts far longer than ibogaine does and gives no number. Averaging two disagreeing figures would produce a value that neither source supports.

Four smaller ones, for completeness. Our secondary file cites the GDNF paper to the wrong journal in its body text and the right one in its bibliography, and presents a study of alcohol consumption in rats as general anti-addiction evidence. It gives the wrong given name for the first author of the 2019 reset proposal. It cites the hERG paper to a journal, volume and page range matching no paper found in checking, while its own bibliography carries the correct one. Our primary file cites a 2018 Mash publication that appears nowhere in its own 21-entry bibliography, so the reference had to be located independently; it is the 191-participant series in section 06. That same file gives the year, volume and pages of the 18-MC review wrongly, although its identifier does resolve to the real paper.

One observation runs the other way and deserves saying as plainly as the rest. Our secondary file is structurally better than the primary on the thing that matters most in this subject: it gives cardiac risk its own numbered Tier-1 section, which the primary file does not. It is also the file carrying more of the errors above. Being more careful about the danger and less careful about the citations is a real combination, and both halves are true of the same document.

So the position is this. Something is being reported at the acute end, by separate groups across the published record, and our own file calls that observational data consistent even as it refuses the data the standing of a result. What has never been done is the single thing that would turn a consistent observation into a result, and both of our files and an independent 2026 review agree it still has not been done. Meanwhile the treatment goes on anyway, mostly in places that exist precisely because the regulated system does not offer it, and the deaths accumulate in a record that nobody can put a denominator underneath. A United States state has now committed public money to running those trials, which is the one thing in this story that changed recently and that neither of our research files has. Whatever those trials find, the sentence that will still be true afterwards is the one that came out of reading the celebrated papers rather than the critics: the studies most often cited to show that ibogaine works each contain, in their own results, a person who did not survive the treatment, and our own research files summarised both of them without mentioning it.