Kratom Effects: What Research Shows – and What Remains Open

"How does kratom work?" is the most frequently asked question about Mitragyna speciosa — and the one that gets the most contradictory answers online. Two experience reports about the same plant often read like descriptions of two entirely different things.

This article answers the question anyway, just not in the expected way. It sets out what has actually been investigated about the plant's constituents, under what conditions those investigations took place, why none of it permits a statement about a human being — and which findings in the literature are solid. The most solid ones, as it happens, concern the risks.

The short answer first

There is no universally valid description of how Mitragyna speciosa works, and that is not caution speaking, it is a lack of data. What exists are three kinds of knowledge whose evidential weight differs sharply:

Preclinical laboratory findings: cell cultures and animal models. Numerous and methodologically sound for the most part — but statements about cells and rodents, not about people.
Self-reports and surveys: extensive, predominantly from the United States. They document what people state; they do not demonstrate an effect.
Documented adverse effects: case reports, poison-control and clinical data. The best-evidenced part of the literature — which is why it is covered here at length.

Controlled clinical trials with a defined quantity of substance, a comparison group and pre-registered endpoints, by contrast, exist only in very small numbers. That gap is precisely why regulatory bodies classify the evidence base as insufficient.

Why the question is so hard to answer

The starting material is not standardised

The content of mitragynine, the principal alkaloid by quantity, varies considerably with origin, tree age, harvest time, drying and storage. Dried leaf typically contains roughly 0.5 to 2 % mitragynine by dry weight. More than 40 alkaloids have been described in total, of which only a handful have been characterised pharmacologically in any detail.

7-Hydroxymitragynine is typically present in raw leaf at below 0.05 % and is additionally formed by oxidation after harvest. It is therefore not a "second active ingredient" of the plant, as is often written, but a minor alkaloid that in humans also arises as a metabolite of mitragynine.

What this means in practice: two samples of identical weight can carry completely different alkaloid profiles. A quantity stated without laboratory analysis is a statement about weight, not about constituents. Comparing two experience reports therefore means, strictly speaking, comparing two unknown preparations.

The controlled comparison is missing

Even with standardised material, a second problem would remain: without a comparison group and without blinding, there is no way to separate what is attributable to the constituents from what is attributable to expectation, daily form or surrounding circumstances. Anyone describing an effect is always describing both at once. Randomised controlled trials were invented for exactly this purpose — and they are largely absent here.

Preclinical and clinical: why this distinction decides everything

Nearly everything known about these alkaloids comes from preclinical research. The term is not a value judgement, it is a statement of location:

Preclinical means cell cultures, isolated receptors, tissue samples, animal models — predominantly mice and rats. What is measured is whether a molecule binds to a target protein, how strongly, and what the cell passes on as a result.
Clinical means investigation in humans, with a defined quantity of substance, a control group, ideally randomised and blinded, with pre-specified endpoints and a published study protocol.

Between the two lies not a small step but the point at which most substance candidates in drug development fail. A finding that is unambiguous in a cell model says nothing about whether the same effect occurs in an organism at all.

From this follows the rule this text observes: a preclinical finding is a statement about a cell model or an animal. It is not a statement about a human being.

Note: this article serves botanical, chemical and legal information purposes. No claims about effects are made. Kratoein products are not intended for consumption and are supplied as collector's and research items.

Which receptor systems have been examined in the laboratory

The findings below come without exception from cell and animal models. They describe molecules at binding sites, not sensations.

µ-Opioid receptor: partial agonist

In binding and functional assays, mitragynine is described as a partial agonist at the µ-opioid receptor. A partial agonist docks and triggers the signal only partially — unlike a full agonist, which elicits the maximum possible response. At the δ- and κ-opioid receptors, the same models report a markedly weaker or antagonistic interaction.

The molecular class is notable: mitragynine is an indole alkaloid and is structurally unrelated to the classical opiates. It is not a morphine derivative. That a chemically entirely different architecture docks at the same binding site is one reason for the scientific interest.

Adrenergic and serotonergic systems

An interaction with adrenergic (particularly α2-) and serotonergic binding sites has also been examined — that is, with receptor systems at which noradrenaline and serotonin respectively dock in the body. These findings are methodologically less consistent than the opioid data, come from different test systems, and have not been independently replicated throughout.

That several receptor systems are involved is itself a finding: it makes any simple "substance X leads to effect Y" assignment untenable, because the contributions of the individual systems cannot be cleanly separated within an organism.

The point hardly any article makes: G-protein bias

After binding, a µ-opioid receptor can set off two different internal pathways: the G-protein pathway and the β-arrestin pathway. In several cell models, mitragynine predominantly activates the G-protein pathway and recruits β-arrestin-2 only weakly — it is described as a G-protein-biased partial agonist.

Why this is pharmacologically interesting: a much-discussed hypothesis linked the β-arrestin pathway to undesirable effects of classical opioids, respiratory depression in particular. Substances that set off one pathway while largely leaving out the other are therefore regarded as templates for developing new analgesics. That is one of the principal reasons why research on these alkaloids is conducted at all.

Equally important is the qualification almost always missing from popular accounts: the bias hypothesis is scientifically contested. More recent work suggests that part of the measured "bias" is simply low intrinsic activity, and that the result varies with test system, receptor density and the analytical model used. The assumed link between β-arrestin-2 and respiratory depression has also not been consistently confirmed in follow-up studies. The finding is therefore an interesting research avenue — not a safety promise, and certainly not a statement about humans.

Why receptor binding permits no inference about a human being

Between a binding curve in a test tube and a human being lie at least five steps of translation. Each one on its own can invalidate the finding.

1. Bioavailability

In cell culture, the substance meets the receptor directly. In an organism it must first be absorbed and pass through the liver. For mitragynine, low oral bioavailability has been measured in animal models: a substantial proportion never reaches the bloodstream at all. A concentration that does something in an assay may never occur in the body.

2. Metabolism

Mitragynine is transformed by enzymes of the cytochrome P450 family, CYP3A4 in particular, alongside CYP2D6 and CYP2C9. Their activity differs genetically from person to person and can be inhibited or accelerated by other substances. Two consequences follow. First, metabolites arise — 7-hydroxymitragynine among them — that are themselves receptor-active, in part more so than the parent compound. Second, this is exactly where the interaction risks with medicines described in the specialist literature are located.

3. Blood-brain barrier

Only what crosses this barrier reaches receptors in the central nervous system. Whether and to what extent a molecule does so depends on lipid solubility, molecular size and active transporters — and cannot be read off a binding study.

4. Dose-response curve

Assays measure at defined concentrations. In an organism the relationship is frequently non-linear; for mitragynine, opposing effects at different concentrations have been described in animal studies. A curve whose shape in humans is unknown permits no statement about any point on it. This is precisely why serious review articles state no threshold values.

5. Species differences

Rodents metabolise alkaloids differently from humans: different enzyme equipment, different half-lives, different metabolite ratios. Added to this is the route of administration — in animal experiments substances are frequently injected, which bypasses hepatic first pass and is not comparable to uptake via the gastrointestinal tract.

Anyone wishing to check the state of publication for themselves will find it in the literature database PubMed.

What surveys show — and what they cannot show

There are extensive self-report surveys, predominantly from the United States, in which participants state widely differing reasons for use. Such surveys are not scientifically worthless: they describe prevalence, patterns and problem areas within a population. But they answer a different question from "what does the substance do?" — namely: "what do people state?"

Four methodological limits are decisive here:

Self-selection: those who voluntarily take part in an online survey about a plant are as a rule not representative. Satisfied participants answer more often than dissatisfied ones, and communities recruit from within themselves.
No placebo arm: without a comparison group there is no yardstick. Expectancy effects are substantial wherever people observe themselves — and cannot be factored out of a survey.
Unknown preparations: what the respondent actually used has virtually never been analysed. Origin, alkaloid profile, contaminants and admixtures all remain open.
Recall bias: retrospective statements about frequency and quantity are systematically imprecise, and events are assembled after the fact into explanations they did not originally have.

The same applies in mirror image to case reports from emergency medicine: they are valuable pointers to possible risks, but very frequently involve the simultaneous use of several substances and have no comparison group. A single case establishes no frequency — in either direction.

Why authorities classify the evidence base as insufficient

That the question is open is not this article's opinion; it matches the assessment of official bodies:

In the EU, Mitragyna speciosa is not regarded as having been consumed as a food to a significant degree before May 1997 and therefore falls under the Novel Food Regulation. Authorisation as a novel food requires a safety assessment reviewed by EFSA; no such authorisation exists.
The German Federal Institute for Risk Assessment (BfR) points in its opinions to insufficient toxicological data and to reports of adverse effects.
The US FDA has approved no use for Mitragyna speciosa and warns of health risks; in 2025 it additionally recommended scheduling the minor alkaloid 7-hydroxymitragynine under controlled-substances law.
The European drugs agency EUDA (formerly EMCDDA) lists the plant as a monitored substance and explicitly describes the evidence base as limited.
The WHO Expert Committee on Drug Dependence concluded in its pre-review that the available data are insufficient for a definitive assessment and recommended continued surveillance.

For readers this means two things: there is no officially recognised application, and there is no officially reviewed quantity. Anyone claiming otherwise in commerce is making an advertising claim — and that is tightly constrained in law. How tightly is set out in the article Kratom and the German Medicines Advertising Act (HWG).

Documented adverse effects

Here the evidence base is comparatively good, for a simple reason: problems get documented, uneventful outcomes do not. That asymmetry has to be kept in mind; it does not, however, make the findings below any less real.

Physical effects

Case series, poison information centres and surveys regularly report nausea and vomiting, constipation, dry mouth, increased sweating, dizziness, headache, itching, as well as tachycardia and changes in blood pressure. Individual cases of seizures and of cholestatic liver injury with jaundice have also been described, which as a rule receded after discontinuation.

Dependence potential and withdrawal

This point is among the best evidenced in the entire literature and is almost universally passed over in sales copy. With regular, longer-term use, tolerance development and a withdrawal syndrome are documented: restlessness, irritability, sleep disturbance, muscle and limb pain, runny nose, sweating, anxiety and pronounced craving. According to the available data, severity and duration relate to how often and how long use continued. Withdrawal signs in newborns following use during pregnancy have also been described. The article Kratom & tolerance goes into this in more depth.

Interactions

Most serious incidents in the literature involve the simultaneous use of several substances. Combinations with depressant substances — benzodiazepines, opioids, alcohol — are named particularly often, as are medicines broken down by the same CYP enzymes. Through inhibition of CYP2D6 and CYP3A4, the breakdown of other active substances can be slowed, so that their levels rise uncontrolled. Details in Kratom & benzodiazepines and Kratom & alcohol.

Contamination and adulteration

A considerable share of the reported harm is attributable not to the plant but to what was sold under its name. Documented are contamination with heavy metals (lead and nickel among others), a larger salmonella outbreak via contaminated goods, and adulteration with synthetic substances: in the best-known case, a synthetic opioid had been mixed into a product sold as kratom, which led to several deaths. Why analysis is not an optional extra is covered in the article Heavy metals & pesticides in kratom.

Conclusion: an honest answer to the opening question

What is reliably known about the alkaloids of Mitragyna speciosa concerns molecules, receptors and cell models. Mitragynine is an indole alkaloid described in preclinical models as a G-protein-biased partial agonist at the µ-opioid receptor; adrenergic and serotonergic interactions are less consistently evidenced. Alkaloid content varies widely, the bias classification is scientifically contested, and between assay and human being lie bioavailability, metabolism, the blood-brain barrier, the shape of the dose-response curve and species differences.

Controlled human data from which a statement about effects could be derived are largely absent — that is the answer to the question, and it is a solid answer, not an evasion. Conversely, the part of the literature describing dependence potential, withdrawal, interactions and contamination is far better evidenced than any narrative about pleasant effects. Anyone taking the research seriously has to take note of that part first.

This text therefore makes no statement whatsoever about what the plant does to a human being. It describes what was measured, where it was measured and what follows from that: little that is certain so far — and a good deal one ought to know before reading something else.

Legal notice

This article is for informational purposes only and does not constitute legal advice. Its content is not intended to encourage consumption. Laws may change; the applicable regulations and information from official bodies are authoritative. Image source: https://www.kratoein.com/

Back to blog