The most common reason a cannabis drug interaction causes harm is not that the interaction is obscure. It is that the patient did not mention the cannabis.
People see cannabis as a herb, or as a personal matter, or as something a doctor will judge them for, so they don’t mention it. Meanwhile the doctor is adjusting a warfarin dose or starting an immunosuppressant without the full picture.
This post puts the main interactions into one reference you can take to an appointment. It does not replace that appointment. Related topics go into more depth: CBD and epilepsy, cannabis and cancer treatment, dosing, how the body processes cannabinoids and who should not use cannabis at all.
This is general education, not medical advice. Do not start, stop or change any medicine without the clinician who prescribes it.
How the interactions work
Two mechanisms account for nearly all of it.
Enzyme inhibition. THC and especially CBD inhibit liver enzymes that clear other drugs. This has been shown mostly in laboratory studies. The enzymes are mainly the cytochrome P450 family (CYP2C9, CYP2C19 and CYP3A4 are the most relevant) and the UDP-glucuronosyltransferase (UGT) enzymes. When you inhibit the enzyme that clears a drug, that drug’s blood level rises. If a medicine has a narrow therapeutic window, a higher level means a higher risk of toxicity.
Enzyme induction (smoked cannabis). Smoke works the other way. Like tobacco smoke, cannabis smoke induces CYP1A2, which speeds up the clearance of drugs that depend on it. This matters most for the antipsychotics clozapine and olanzapine (see below).
Additive pharmacodynamics. THC is sedating and affects heart rate and blood pressure. Taken with other sedatives or with heart and blood-pressure drugs, the effects add up. No enzyme is involved, just two drugs doing the same thing.
The enzyme effects are stronger with CBD at high doses than with occasional low-dose THC. Pharmaceutical CBD for epilepsy is dosed at up to 20 mg/kg/day (25 mg/kg/day for TSC). For a 60 kg adult, 20 mg/kg/day is 1,200 mg a day. That is a very different pharmacological situation from a wellness product containing 25 mg. A 2019 review of CBD’s adverse effects and toxicity (Huestis et al.) found drug–drug interactions and liver abnormalities reported in human studies of epilepsy and psychiatric patients. It concluded that CBD is not risk-free.
The interaction with the strongest evidence: warfarin
A systematic review (Lopera, Rodríguez & Amariles, 2022) looked at a decade of literature and found 19 drug–cannabis interaction pairs. It ranked them by risk level: 1 (very high), 2 (high), 3 (medium) and no risk. Only one pair reached level 1.
Cannabis and warfarin. Four case reports documented INR values rising from a target of 2–3 to an average of about 7.6, with gastrointestinal bleeding. The likely mechanism is cannabinoid inhibition of CYP2C9, the main enzyme that clears warfarin. One earlier case (Yamreudeewong et al., 2009) involved a 56-year-old man on long-term warfarin who was admitted with upper gastrointestinal bleeding and a sharply raised INR while smoking cannabis. The authors rated the interaction as “probable”.
The reviewers’ recommendation is that the combination could be contraindicated. If the two are used together, the warfarin dose should be adjusted and INR monitored closely.
If you take warfarin and you use cannabis in any form, this is the single most important sentence in this post: tell the clinician managing your anticoagulation. They need to decide whether to adjust the dose, check your INR more often, or advise against the combination. A bleed is a preventable event.
Direct oral anticoagulants such as rivaroxaban and apixaban are partly cleared by CYP3A4. The concern about the mechanism is the same, but there is very little clinical data.
The high-risk pairs
Two pairs were placed at level 2 in the same review. Both showed raised drug levels, without clinical-trial evidence to define the risk:
Buprenorphine. Levels increase. In one retrospective analysis, buprenorphine concentrations were higher in cannabis users. Buprenorphine is cleared by CYP3A4 and UGT enzymes, and CBD inhibits both in laboratory studies. The reviewers advise avoiding the combination. If it cannot be avoided, the buprenorphine dose should be adjusted and levels monitored.
Tacrolimus. This immunosuppressant is given after organ transplantation. It has a narrow therapeutic window, with serious consequences at both ends: too high damages the kidneys, too low risks rejection. Case reports describe large increases in tacrolimus levels with CBD. The recommendation is the same as for buprenorphine. For transplant patients, talking to the transplant team is not optional. Ciclosporin and sirolimus share the CYP3A4 route and deserve the same caution.
The medium-risk group
Twelve of the nineteen pairs (63%) were rated medium risk. Most of them are nervous-system medicines: clozapine, methadone, clobazam, chlorpromazine, eslicarbazepine, hexobarbital, stiripentol and valproate. The group also includes indinavir, ketoconazole, rifampicin and theophylline. Three deserve detail:
Clobazam. This is the best-documented cannabinoid interaction, because it was measured during pharmaceutical CBD (Epidiolex) development. CBD inhibits CYP2C19 and raises levels of clobazam’s active metabolite, N-desmethylclobazam. The Epidiolex prescribing information reports roughly a 3-fold increase. It advises considering a lower clobazam dose if clobazam side effects such as sleepiness appear. Doctors usually manage this with dose adjustment rather than by avoiding the combination. Non-pharmaceutical CBD products would likely act in the same direction, but the size of the effect depends on dose and product.
Valproate. The problem here is not raised valproate levels. It is liver enzyme (transaminase) elevation. The Epidiolex label states that most ALT elevations in its controlled studies occurred in patients also taking valproate, so liver tests are monitored when the two are combined. One trial also reported low platelet counts in patients taking both, but the finding was not statistically significant. Ask whether platelet checks are needed.
Clozapine and olanzapine. Smoked cannabis induces CYP1A2, so a regular smoker may run lower levels of these antipsychotics. The real danger comes when someone stops smoking: the induction wears off and levels can rise sharply, with a risk of toxicity. Separately, high-THC cannabis can worsen psychotic symptoms and work against the medication.
Summary table
| Drug / class | Examples | What can happen | Mechanism | Risk | Evidence |
|---|---|---|---|---|---|
| Vitamin K antagonists | Warfarin | ↑ INR, bleeding | CYP2C9 inhibition | Very high (review level 1) | Case reports |
| DOACs | Rivaroxaban, apixaban | Possible ↑ levels | CYP3A4 inhibition | Uncertain | Mechanistic only |
| Buprenorphine | — | ↑ levels | CYP3A4 + UGT inhibition (lab data) | High (level 2) | Retrospective + case report |
| Tacrolimus, ciclosporin | — | ↑ levels, toxicity | CYP3A4 inhibition | High (tacrolimus level 2) | Case reports + small trial |
| Other opioids | Fentanyl, oxycodone | Extra sedation; possible ↑ levels | Pharmacodynamic + CYP3A4 | High | PK + observational |
| Morphine | — | Extra sedation | Pharmacodynamic only | Moderate | Mechanistic |
| Benzodiazepines, Z-drugs | Diazepam, alprazolam | Extra sedation; possible ↑ levels | Pharmacodynamic + CYP3A4/2C19 | High | PK + mechanistic |
| Alcohol | — | More impairment, including driving | Pharmacodynamic | High | Experimental human studies |
| Clobazam | — | ↑ N-desmethylclobazam (~3-fold), sleepiness | CYP2C19 inhibition (CBD) | Medium (level 3); well characterised | Clinical trials, label |
| Valproate | — | ↑ liver enzymes | Additive liver effect (transaminase elevation) | Medium (level 3) | Clinical trials, label |
| Clozapine, olanzapine | — | ↓ levels while smoking; ↑ when smoking stops; worse psychosis | CYP1A2 induction by smoke + CB1 | Medium (clozapine level 3); high in psychosis | PK + epidemiological |
| Phenytoin | — | Possible ↑ levels | CYP2C9 (theoretical) | Uncertain | Mechanistic |
| Carbamazepine | — | ↓ cannabinoid levels | CYP3A4 induction by carbamazepine | Moderate | PK/mechanistic |
| Calcium channel blockers | Amlodipine, diltiazem | Possible ↑ levels; extra BP drop | CYP3A4 + pharmacodynamic | Moderate | Mechanistic |
| Beta-blockers | Metoprolol, atenolol | THC tachycardia may partly offset effect | Pharmacodynamic | Possible, limited data | Mechanistic |
| SSRIs | Citalopram, sertraline | Possible ↑ levels; extra sedation | CYP2C19 (theoretical) | Low–moderate | Theoretical; no controlled data |
| Sulfonylureas | Glipizide, glyburide | Possible ↑ levels, low blood sugar | CYP2C9 (theoretical) | Moderate | Mechanistic |
| Chemotherapy, TKIs | Paclitaxel, docetaxel, irinotecan, imatinib | Possible ↑ toxicity | CYP3A4 inhibition | High if present | Mechanistic; no trials |
| HIV protease inhibitors | Indinavir | Altered levels | CYP3A4 | Medium (level 3) | Clinical study |
The classes to raise, by category
Anticoagulants and antiplatelets. Warfarin above all. DOACs too, where the data is thin.
Antiepileptics. Clobazam, valproate, stiripentol, phenytoin and carbamazepine.
Sedatives and CNS depressants. Benzodiazepines, Z-drugs, opioids, some antihistamines and alcohol. The sedation adds up, and this combination does not need an enzyme mechanism to be dangerous. It is especially risky for older people and anyone with sleep apnoea.
Immunosuppressants. Tacrolimus, ciclosporin and sirolimus. Narrow windows, serious consequences.
Cardiovascular drugs. THC raises heart rate, and blood pressure can drop on standing (orthostatic hypotension), which increases fall risk in older people. Take care when combining THC with antihypertensives or beta-blockers, or if you already have heart disease. Both dronabinol and nabilone carry cardiac warnings in their labelling.
Antiretrovirals. Many are cleared by CYP3A4, and indinavir appears in the review’s medium-risk list.
Chemotherapy agents. Many are CYP3A4 substrates, including taxanes, irinotecan and tyrosine kinase inhibitors. Nausea control is one of the most common reasons cancer patients use cannabis, so this class deserves specific attention rather than assumptions.
Psychiatric medications. SSRIs, antipsychotics and lithium. The mechanisms vary, and the sedative and psychoactive effects add up.
Diabetes medicines. Sulfonylureas, in theory, through CYP2C9. THC can also affect appetite and food intake, so check blood glucose more often after any change.
If you have liver disease, the risk of drug build-up is higher because the same liver enzymes are involved.
What the evidence cannot show
Being clear about what the evidence cannot show, because overstating the risk also causes harm:
- The systematic review included only 26 articles from ten years of literature. That is a thin evidence base for such a common question.
- Four of the nineteen pairs were rated no risk.
- Most interactions are documented as pharmacokinetic changes (a measured shift in drug levels) rather than proven clinical harm. A change in levels is a reason to monitor, not automatically a reason to stop.
- Most evidence comes from laboratory enzyme studies, case reports, and pharmaceutical CBD trials at fixed doses. Results from those trials do not translate directly to other products.
- The size of each effect is uncertain. For example, the Epidiolex label reports that CBD did not change exposure to tolbutamide, a CYP2C9 test drug, yet case reports still link cannabis to raised INR with warfarin.
- The reviewers called the available information limited and said clinical trials are needed.
For most pairs, monitoring is enough. For warfarin and tacrolimus, the decision to continue, adjust or stop belongs to the treating clinician. The same applies to buprenorphine, the review’s other high-risk pair.
How to have the conversation
Say it plainly, and say it first. “I use cannabis” is more useful than a doctor discovering it from an unexplained lab result three months later.
Bring specifics. The form: flower, oil, edible, or prescribed product. The approximate amount and how often. Whether it is THC-dominant, CBD-dominant or balanced. These details change the answer.
Ask three questions:
- Does anything I take get cleared by CYP2C9, CYP2C19, CYP3A4 or UGT?
- Does anything I take have a narrow therapeutic window?
- Should we monitor anything more often?
Mention any change in pattern. Starting, stopping or substantially changing your dose all shift the enzyme effects, and stopping is as much a change as starting. A patient whose warfarin dose was stabilised while using cannabis daily will see their INR move if they stop. A clozapine patient who stops smoking can see their levels rise.
If you are in an emergency department, say it out loud. Standard drug screens do not test for everything, and clinicians make decisions in minutes.
The Thai context
As of October 2026, cannabis flower in Thailand is a controlled herb under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis) B.E. 2568, issued 26 June 2025. The rules are:
- Flower may be dispensed only with a prescription on form ภ.ท. 33 (PT 33) from an authorised practitioner.
- Each prescription covers no more than 30 days and can be used once (no refills).
- Flower that is sold must come from a GACP-certified cultivation site.
- Flower may not be sold or advertised online, and recreational use is not allowed.
Published legal analyses list seven categories of practitioner who can issue the prescription: medical doctors, Thai traditional medicine practitioners, applied Thai traditional medicine practitioners, folk healers, Chinese medicine practitioners, pharmacists and dentists. If you are unsure, ask the practitioner whether they are authorised to issue a PT 33.
Two Ministerial Regulations of B.E. 2569 tightened licensing in 2026 but did not change the PT 33 rules. The Cannabis Control Act is still a draft: the Cabinet approved it on 22 September 2026 and sent it to the House. For later changes, see the 2026 Ministerial Regulations and the PT 33 enforcement update. If you are travelling on from Thailand, note that most neighbouring countries prohibit cannabis.
This has one specific consequence for interactions. The person who prescribes your cannabis and the person who prescribes your other medicines may be different people, in different systems, who do not see each other’s records. A folk healer issuing a PT 33 may have no view of a hospital cardiologist’s warfarin dosing, and the other way round. The patient may be the only person who sees the whole list.
So carry the list. Write down everything you take, and show the same piece of paper to the PT 33 prescriber, your hospital doctor and the Thai pharmacist who dispenses your medicines. Include:
- prescription medicines;
- Thai traditional medicine formulas that contain cannabis;
- cannabis oils from a hospital or clinic;
- CBD products;
- supplements and herbal products.
The short version
The interaction with the best evidence is warfarin, and it is serious. Buprenorphine and tacrolimus come next. Clobazam and valproate are well understood and are managed with monitoring. Clozapine and olanzapine levels can jump when a smoker stops. Antiepileptics, sedatives, immunosuppressants, heart and blood-pressure drugs, antiretrovirals and chemotherapy agents are the classes to raise. For most pairs, monitoring is enough. For the high-risk pairs, your clinician decides.
The failure mode is silence, and it is entirely avoidable.
Sources:
- Lopera V, Rodríguez A, Amariles P. Clinical Relevance of Drug Interactions with Cannabis: A Systematic Review. J Clin Med 2022;11(5):1154.
- Alsherbiny MA, Li CG. Medicinal Cannabis—Potential Drug Interactions. Medicines (Basel) 2019;6(1):3.
- EPIDIOLEX (cannabidiol) oral solution, U.S. prescribing information, revised 5/2026: sections 2 (dosing), 5.1 (transaminase elevations with valproate), 7.2 (clobazam) and 12.3 (pharmacokinetics).
- Huestis MA et al. Cannabidiol Adverse Effects and Toxicity. Curr Neuropharmacol 2019;17(10):974–989.
- Yamreudeewong W et al. Probable interaction between warfarin and marijuana smoking. Ann Pharmacother 2009;43(7):1347–1353.
- Stout SM, Cimino NM. Exogenous cannabinoids as substrates, inhibitors, and inducers of human drug metabolizing enzymes. Drug Metab Rev 2014;46(1):86–95.
- “Dronabinol”, StatPearls, NCBI Bookshelf.
- Cesamet (nabilone) FDA prescribing information.
- Tilleke & Gibbins analysis of the Notification on Controlled Herbs (Cannabis) B.E. 2568.