When Ibogaine Does Not Work

Every clinic that sells this treatment can point to someone it worked for. The harder question is what happened to everyone else, and the published record on that is smaller than almost anyone assumes.

The whole follow-up record

Four studies are usually cited. They are not four independent bodies of evidence.

The last two are the same people. What exists is two small prospective cohorts and one survey of the reachable survivors of a single clinic.

The effect decays, and the decay is in the data

The New Zealand and Mexican cohorts both followed people for a year, and both found the same shape.

In the Mexican cohort of thirty, improvement in drug use was greatest at one month and, in the authors’ own words, never again reached equivalence to that first-month effect. Twelve of the thirty met their definition of a favourable outcome at nine or twelve months.

The attrition matters more than the percentages.

The series moves up and down, from six at six months to eleven at nine months, because which people could be reached changed from one round to the next. The fourteen reached at six months are not the fourteen reached at twelve.

What the missing data change is narrower than it first appears, and worth stating precisely. Compared with where people started, the improvement in drug use scores stayed significant at every time point. What does not survive the adjustment is the claim that the later months held up to the standard of the first: on that test, the authors report no significant result at any point once missing data are accounted for. Their own reading is that a subset of people responded and kept responding, rather than that the group as a whole did.

Three of the thirty went back for another ibogaine treatment within the year.

The New Zealand cohort tells the same story from a different angle. Self-reported opioid use in the previous thirty days stood at 43 per cent at three months and 50 per cent at six, of fourteen people each time, and 45 per cent at twelve months, of the eleven still in the study. Roughly half were using again at every point measured, including the last.

Urine testing, available for only part of the cohort, ran lower than what people said: one positive in eight at three months, one in seven at six, two in eight at twelve. The two series disagree, and the paper says the self-reports exceeded the urine findings rather than treating either as settled. The headline improvement in drug use scores was real and statistically significant, and it rests on eight people who completed every interview.

Of the three participants who left the study, all three had relapsed. Attrition here did not run at random. It ran toward the better-looking result.

One of the seven outcome measures moved in the wrong direction: the medical composite rose significantly over the year, from zero at baseline. The authors read that as either more health problems or more willingness to report them.

Seventy per cent, and where the number comes from

The figure everyone quotes is from a survey of 88 people treated at one clinic in Mexico between 2012 and 2015. Thirty per cent said they never used opioids again. Forty-one per cent were in sustained abstinence of more than six months when surveyed. Seventy per cent had relapsed at some point.

Those three are all percentages of the 88, and they are the ones worth quoting. Two others circulate widely and are almost always quoted wrongly.

The denominator above the sample matters too. The clinic had treated 336 people in that window. 285 could be contacted, 134 began the survey, and 88 remained after exclusions. That is 26 per cent of the treated group. Anyone who had died, who was too unwell to answer, or who had no wish to hear from the clinic again is absent from the result by construction, and the authors say plainly that this may have inflated their estimates of treatment response.

Three of the five authors worked for the clinic being studied, its owner among them. The clinic also paid for the ethics application and the participant incentives, and paid the lead author to design the survey. All of that is disclosed in the paper, and none of it makes the numbers wrong. It does mean the study cannot be read as independent evaluation.

The hardest part comes later

The most useful finding in this literature is the least quoted. A re-analysis of 73 of those same respondents asked what the greatest challenge of the treatment had been. Sixty-one per cent of everything people described fell after treatment rather than during it. The single commonest theme was integration, meaning the difficulty of fitting what had happened into an ordinary life.

It accounted for 23 per cent of the statements made by people the study classed as treatment responders, and 21 per cent of those made by non-responders. Whatever distinguishes a success from a failure here, it is not whether the aftermath was hard.

The other post-treatment themes were health problems, emotional difficulty, fatigue and cost.

Read this carefully: these are counts of statements, not of people, and the non-responder group is fourteen people producing twenty-nine statements. The direction is clear. The proportions are not precise.

The same 88 people, two headlines

The 2017 paper on this cohort reports that 70 per cent relapsed. The 2018 paper on the same cohort reports that 59 of 73 were treatment responders, which is 81 per cent.

Both are true. A responder is someone abstinent or using less than before. A relapse is any return to use at any point. A clinic quoting the second number and a critic quoting the first are describing the same eighty-eight people.

What the trials have not answered

Two randomised controlled trials of ibogaine have completed.

Neither has posted results on the registry, and neither has appeared in the peer-reviewed literature. The larger trial’s registry entry was last updated in August 2024.

Even if results appear, the larger trial will not answer this page’s question. Its primary outcome was withdrawal severity averaged over days two to six. That is whether ibogaine gets someone through detoxification, which is the part of the claim least in dispute. Whether they use again three months later was not measured.

There is, at the time of writing, no completed randomised controlled trial of ibogaine for substance dependence reporting drug use outcomes. Everything above comes from observational work.

What the biggest series does not tell you

The largest medically monitored case series, 191 patients treated in St Kitts, is frequently cited as evidence that ibogaine works. It reports no deaths and no serious adverse events, which is a meaningful safety observation.

It reports no drug use outcome at all. What it measured was mood, craving and physician-rated withdrawal, and at one month those measures existed for roughly a third of the 191, falling to between 28 and 37 people on any single instrument. The paper’s own author declares that she is a founder and shareholder in a company developing ibogaine and holds patents on its active metabolite. The disclosure is in the paper.

A study can be honest, disclosed, and still not answer the question it is quoted for.

If it did not work for you

Three things in the record are worth knowing.

The first is that you are not unusual. On the only numbers that exist, returning to use is the commonest outcome, and about half of one monitored cohort was using again at every point over a year.

The second is that the period people found hardest was the one after they went home, and that this was as true of the people the studies counted as successes.

The third is that leaving an opioid agonist to try this carries its own risk. If tolerance falls and use resumes, the dose that was survivable before may not be. Ibogaine and the heart covers the acute risk; this one belongs to the weeks afterwards.

Nothing on this page is treatment advice, and this site has no clinic to send anyone to. It exists because the studies above are public and the marketing is louder than they are.