CBD Drug Interactions: One Enzyme, One Case File, and the Gap Between a Gummy and the Evidence

Type CBD drug interactions into a search box and the page that comes back is a list: every cytochrome P450 enzyme the liver owns, followed by every drug class that uses one. Steroids, statins, calcium channel blockers, antihistamines, antivirals, benzodiazepines, antidepressants, antiepileptics, beta blockers, proton pump inhibitors, NSAIDs, sulfonylureas. That is a list of medicine. It tells a person on sertraline and a person on warfarin the same thing, which is nothing, and it leaves out the one detail that decides whether an interaction exists, which is the milligrams.

The story with a drug label behind it is small and specific. One enzyme has a measured, labeled effect: CBD is a moderate inhibitor of CYP2C19, by the FDA label for Epidiolex, the prescription form of the molecule. One drug has a patient-level case file: warfarin, seven published cases, two of them with the CBD dose known. Two interactions were measured in dedicated trials and printed on that label: clobazam, whose active metabolite roughly triples, and valproate, which raises the rate of liver enzyme elevations when the two are taken together. A third, tacrolimus, roughly tripled in a controlled 2025 trial, and the Prograf label carries its own cannabidiol section. Between the gummy in your hand and the doses at which most of that was measured sits a gap of ten to seventy-five times.

This page is the map. Which drug class, which enzyme its dosing rules run on, what the label or a trial showed, how strong the evidence is, and which EdibleRank guide holds the detail. It also says what has been measured at a gummy dose, which until 2026 was nothing, and what still has not been.

Which liver enzyme does CBD actually inhibit?

CYP2C19, moderately, by the Epidiolex label. The label's CYP3A4 probe, midazolam, did not move at 750mg of CBD twice daily, its CYP2C9 probe, tolbutamide, did not move at 7.5 mg/kg twice daily, and CYP2D6 is not named anywhere on it. CBD also weakly inhibits CYP1A2 and UGT1A9 and raises levels of some P-gp drugs.

The label is the one place a regulator made a manufacturer run the standard probes and print the answers. Section 12.3 reads like a receipt: caffeine, the CYP1A2 probe, 95 percent more total exposure at 750mg twice daily; bupropion, the CYP2B6 probe, 20 percent less; tolbutamide unchanged; midazolam unchanged; mycophenolic acid, the UGT1A9 probe, 35 percent more; everolimus, a P-gp drug, 2.5 times more at 12.5 mg/kg twice daily. Section 7 turns those into instructions. For CYP2C19 drugs, consider lowering the dose. For clopidogrel, a prodrug that CYP2C19 has to switch on, consider raising it. Rifampin, a strong inducer of both CYP3A4 and CYP2C19, cut CBD exposure 32 percent and its active metabolite 63 percent, so the label allows up to a doubling of the Epidiolex dose alongside one. Strong inhibitors, itraconazole and fluconazole, changed little. That is the entire enzyme content of the only CBD label in existence, and it is shorter than most blog posts on the subject.

The second measurement came out of a brownie. Bansal and colleagues at the University of Washington and Johns Hopkins fed 18 healthy adults a brownie carrying 640mg of CBD and 20mg of THC, then a cocktail of five probe drugs. Omeprazole, the CYP2C19 probe, showed 207 percent more exposure than after a placebo brownie. Losartan, the CYP2C9 probe, 77 percent more. Midazolam 56 percent more, caffeine 39 percent, and dextromethorphan, the CYP2D6 probe, unchanged. A brownie with 20mg of THC and no CBD inhibited nothing at all. So the brownie agrees with the label on the enzyme that matters most and the enzyme that does not, and disagrees on CYP2C9 and CYP3A4, where a single prescription-grade probe found nothing and a single edible found something. Doohan's group in Sydney, working in a dish with twelve cannabinoids, found nearly all of them inhibiting CYP2C9 at clinically relevant concentrations and leaving CYP2D6, CYP3A4 and CYP2B6 mostly alone.

Three experiments, one enzyme they agree on. The honest summary is that CYP2C19 is the CBD enzyme, CYP2C9 is contested between a label probe that saw nothing and a brownie that saw 77 percent, CYP3A4 runs from nothing to modest, and CYP2D6 is nothing. Tacrolimus is the warning against reading too much into that summary: it runs on CYP3A and P-gp, and it tripled at a dose where the label's CYP3A4 probe predicts nothing, by a route nobody has pinned down. The internet's list treats all four as equal, which is how a beta blocker ends up on the same warning as clobazam.

Which medications does CBD interact with?

The ones with evidence are warfarin (case reports with INR readings), clobazam (its active metabolite roughly triples), valproate (more liver enzyme elevations), citalopram and escitalopram (levels up in two studies at hundreds of milligrams a day), tacrolimus and everolimus (exposure up about threefold and 2.5-fold), and clopidogrel (a label inference, in the direction of less effect). Everything else on the internet's list is enzyme theory.

The table is the page. Each row names the drug class, the enzyme its own dosing rules run on, what the Epidiolex label or a human study showed, how strong that evidence is, and where on EdibleRank the detail lives.

The warfarin row is the one with a human being in it. Grayson's patient at the University of Alabama at Birmingham was a 44-year-old with a mechanical mitral valve, stable on 7.5mg of warfarin a day, enrolled in the compassionate-use Epidiolex program. At 265mg of CBD twice daily his INR was 2.96. Two weeks later, at 528mg twice daily, it was 6.86, more than twice the top of his target range, and over the following months his warfarin came down to 5.36mg to hold him in it. Cortopassi's patient needed a cut of nearly 20 percent. Smythe's systematic review found seven warfarin cases in total and nothing, anywhere, for any other anticoagulant; the doses involved were either above 600mg of CBD a day or unknown. The case authors blame CYP2C9. The label's CYP2C9 probe saw nothing. The INR moved anyway, which is the difference between a mechanism and a measurement, and the blood thinners guide holds all seven cases.

The clobazam row is the one with the cleanest number. Geffrey's team at Massachusetts General Hospital watched 13 children on clobazam start CBD: the active metabolite, N-desmethylclobazam, rose by a mean of 500 percent at four weeks, clobazam doses were cut in 10 of the 13, and side effects eased once they were. Morrison then ran the healthy-volunteer version for the manufacturer and got 3.4-fold, and the label rounds that to approximately threefold at 750mg twice daily. The label also prices it: somnolence and sedation in 46 percent of patients on both drugs against 16 percent of those on CBD without clobazam. Our benzodiazepines guide carries the sedation arithmetic for alprazolam and lorazepam, which turn out to be a different question, since neither runs on CYP2C19.

The valproate row is a liver finding rather than a blood-level finding. Morrison found CBD had no clinically relevant effect on valproate exposure, and the label agrees. What the label reports instead, in its hepatic injury warning, is that ALT rose past three times the upper limit of normal in 21 percent of patients taking valproate alongside Epidiolex, 30 percent in those also on clobazam, and 3 percent in patients on neither. Gaston's UAB program, 81 patients titrated from 5 to 50 mg/kg a day, saw the same abnormal liver tests cluster in the valproate group. Two drugs that each ask something of the liver, asked at once. Our epilepsy page covers the trials these numbers came from.

Does THC have the same interactions as CBD?

Mostly the reverse. THC is cleared by CYP2C9 and CYP3A4, and in the brownie study the THC-only brownie inhibited none of the five probe enzymes. What the literature shows is other drugs changing THC: ketoconazole raised peak 11-hydroxy-THC 204 percent and rifampicin cut it 87 percent, and 640mg of CBD raised THC exposure 161 percent.

Stott's 2013 study for GW Pharma gave 36 healthy men four sprays of Sativex, 10.8mg of THC with 10mg of CBD, alone and after one of three drugs. Rifampicin, the inducer, dropped peak THC 36 percent, peak CBD 52 percent and peak 11-hydroxy-THC 87 percent. Ketoconazole, the CYP3A4 inhibitor, raised them 27, 89 and 204 percent. Omeprazole, a CYP2C19 inhibitor, changed nothing. Our 11-hydroxy-THC explainer owns the enzyme facts for THC and what they do to the size of an edible high; the point for this page is that THC is the molecule that gets changed, and CBD is the molecule that does the changing. Bansal's brownie made the same point inside one experiment: 640mg of CBD raised the exposure to the 20mg of THC riding alongside it by 161 percent, which the authors read as CBD blocking the CYP2C9 clearance of oral THC.

THC's documented interactions at gummy doses are additive effects rather than blood-level changes: sedation on top of a benzodiazepine, mirtazapine or trazodone, impairment on top of alcohol, more propofol needed in the operating room. Those live on the sibling guides, including the surgery guide, because the enzyme map has nothing to say about them. The THC warfarin cases, five of them with doses mostly unknown, are the exception, and the blood thinners guide lists each one.

How much CBD is in a gummy compared with the doses in these studies?

Ten to seventy-five times less. Wyld's Strawberry 20:1 carries 20mg of CBD per gummy by Wyld's own spec; Kanha's 4:1 Peach carries 10mg. The citalopram, clobazam and brownie studies ran at 200 to 1,500mg of CBD a day, so matching them takes 10 to 75 Strawberry gummies or 20 to 150 Peach ones.

Wyld's product page puts the Strawberry at 20mg of CBD and 1mg of THC per gummy, 200mg and 10mg per container of ten, with the note that packaging and dosing vary by state. A whole container, eaten at once, lands at the bottom of Anderson's 200 to 800mg range and under everything else on this page. Reaching Bansal's brownie takes 32 gummies, and the label's 750mg twice daily takes 75. Kanha's spec sheet gives the 4:1 Peach 10mg of CBD and 2.5mg of THC per piece, 100mg and 25mg per bag; the brownie dose by that route is 64 gummies, which also delivers 160mg of THC, against the 20mg Bansal's brownie carried. The experiment would end for other reasons.

There is a quiet comedy in a warning the label earned at seventy-five gummies a day being repeated to someone eating one, and a quieter one in the fact that the warning still stands, for a reason the hemp aisle cannot explain: almost nobody has looked at the doses in between. The only published reading inside the gummy range for warfarin is Thomas's 85-year-old on 5.3mg of CBD and 0.3mg of THC a day, whose INR barely moved across eleven months. One patient, one negative, and a quarter of a Strawberry gummy.

CBD oil by the dropper is the exception, as on every guide in this cluster. A few hundred milligrams a day, which the hemp market recommends for pain and sleep, sits inside the range where Anderson measured citalopram levels rising and where Smythe found the warfarin cases cluster. The product format decides which half of this page applies.

Has anyone studied a gummy-sized dose of CBD with a prescription drug?

Once, in 2026. Gorbenko and colleagues gave twelve healthy volunteers a single 30mg dose of CBD an hour before amitriptyline and tramadol: amitriptyline exposure rose 13 percent and tramadol did not change. No study has measured a nightly gummy against warfarin, clobazam, citalopram or valproate, by a PubMed search on November 7, 2026.

The Leiden study is small and its number is small. Thirteen volunteers enrolled, twelve finished, and 30mg of CBD taken fasting raised the total exposure to a 25mg amitriptyline dose by 13 percent and the peak by 17 percent, with nortriptyline, tramadol and tramadol's active metabolite unchanged. The authors expect chronic dosing and food to enlarge the effect, and the label supports the second half of that: a high-fat meal raised peak CBD five-fold and total exposure four-fold in healthy subjects. A gummy eaten after dinner is not the same dose as the same gummy eaten on an empty stomach, which no tin mentions either. Set 13 percent next to the clobazam metabolite's threefold and the citalopram study's 43 percent and the gummy dose is a different order of magnitude, measured once, against one drug.

Warren's 2026 review of low-dose CBD, defined as 175mg a day or less, puts the pharmacology behind that gap. Across the pharmacokinetic studies, peak and total CBD exposure ran 5 to 100 times below a 20 mg/kg Epidiolex dose. Interactions were observed at 10 to 100mg with THC, amitriptyline and hydromorphone. No serious adverse events were reported at any low dose. And, the finding the hemp aisle will not print, the randomized trials showed little evidence that low-dose CBD did anything for pain, fibromyalgia, stress, Parkinson's or Crohn's either. A molecule that does little at 25mg is also a molecule that interacts little at 25mg. The two facts arrive together, and the industry has chosen to advertise only the second.

What remains unmeasured is exactly the question people ask. On the patient.info forum in July 2019, a poster on 10mg of propranolol and 100mg of levothyroxine described buying hemp oil capsules on Amazon with the idea of replacing the propranolol, and asked whether the two could be taken hours apart. The one reply, twelve days later, was the Indiana University list of drug classes that use CYP450, nineteen of them from steroids to sulfonylureas with beta blockers starred, plus the advice to take one or the other. Nobody in the thread mentioned a milligram of CBD, and the capsule's contents never came up. That exchange is the state of the art for a person on a beta blocker, and the enzyme map above has no row for propranolol because a PubMed search on November 7, 2026 turned up no study to put in one.

What the prescriber can use

Vague disclosure gives a prescriber nothing to check against the table above. These four details do:

1. The product and the CBD in milligrams per day. Twenty milligrams and 600mg are different pages of this guide.
2. The THC in milligrams per day, since THC is the half that stacks sedation and the half that other drugs change.
3. Which row of the table the prescription sits in, because the follow-up differs by row: an INR for warfarin, a metabolite level for clobazam, a liver panel for valproate, side effects for an SSRI.
4. Whether the CBD is a gummy or a dropper, since the dropper is the format that reaches the doses in the studies.

How to bring cannabis up with a doctor is covered in our guide to using edibles for medical purposes.