In 2014, a study in JAMA Internal Medicine found that states with medical cannabis laws had a 24.8% lower opioid overdose death rate than states without them, and the cannabis industry has been quoting it ever since. Five years later a Stanford team ran the same analysis with data through 2017, and the association flipped: medical cannabis states went from 21% lower overdose mortality to 23% higher. The authors called the original link most likely spurious. The talking point has outlived its data by about seven years.
People searching for cannabis and opioids usually mean one of two different things. One is withdrawal, the miserable days after stopping. The other is treatment, whether cannabis can stand in for methadone or buprenorphine. Our position on both is firm. The withdrawal evidence is a prescription THC capsule given in a hospital, with a modest effect and a racing heart at the doses that worked. And as a replacement for medication, cannabis has nothing that compares with the drop in overdose deaths that methadone and buprenorphine show. This page covers opioid use disorder as an addiction. Opioids prescribed for pain, and cannabis as a way to need fewer of them, are on our chronic pain page.
Did medical cannabis laws lower opioid overdose deaths?
The first analysis said yes and the longer one said no. Bachhuber 2014 compared state death certificates from 1999 to 2010 and found 24.8% lower opioid analgesic overdose mortality in medical cannabis states. Shover 2019 used the same methods through 2017, and the association reversed to 23% higher, even after accounting for recreational laws.
Neither paper measured whether any individual swapped a pill for a joint. Both are ecologic studies, comparing state totals, and the PNAS paper notes that the original authors had urged caution about drawing individual conclusions from them. The PNAS team made the arithmetic point bluntly: medical cannabis is used by about 2.5% of the US population, and a group that small is unlikely to push national overdose deaths down in one decade and up in the next. They read both results as noise from aggregate data, and concluded that the claim that passing a medical cannabis law will reduce overdose deaths deserves skepticism. A prospective look at individuals, using the national NESARC survey, ran the other way too: adults who used cannabis in 2001 to 2002 had higher odds of developing prescription opioid use disorder three years later (adjusted odds ratio 2.18).
Can cannabis help with opioid withdrawal?
Modestly, in two inpatient studies of dronabinol, a prescription synthetic THC capsule. Bisaga 2015 randomized 60 people in detox at Columbia to 30mg a day or placebo, and withdrawal scores were lower on dronabinol. It did nothing for the outcome that matters afterward: 35% in each group completed the eight-week trial.
The Columbia design is worth reading closely. Forty people got dronabinol and twenty got placebo while they went through inpatient detox, took an injection of extended-release naltrexone before discharge and a second one four weeks later, and kept taking the capsules for five weeks at home. Successful naltrexone starts ran 66% on dronabinol and 55% on placebo, a gap the trial could not distinguish from chance. The curious finding was a post hoc one. The 32% of participants who chose to smoke marijuana regularly during the outpatient phase reported less insomnia and anxiety and were more likely to finish, whichever capsule they were on. That is an observation about a self-selected group, and the authors reported it as one.
The Kentucky group under Sharon Walsh then tested dronabinol head to head in 12 adults kept on oxycodone 30mg four times a day, switched to placebo for 21 hours before each session to bring on withdrawal. At 5 and 10mg, dronabinol looked like placebo. At 20 and 30mg it produced modest withdrawal relief along with more high, sedation, bad effects and a racing heart. The planned 40mg dose caused sustained tachycardia with anxiety and panic, and was cut to 30. The authors concluded that dronabinol is not a likely candidate for withdrawal treatment.
For scale, 20 to 30mg sits in the high tier (15 to 30mg) on our dosing guide. The doses that did anything are the doses where a first-time edible user is likely to have a bad night, given in these trials to people already in withdrawal and under inpatient observation. An edible adds its own timing problem: onset runs 30 minutes to 2 hours after eating and the main effects last 4 to 8 hours, per our edible timing guide. Withdrawal symptoms come and go on their own schedule, and a gummy cannot be steered against them.
What changes when the opioid is fentanyl?
None of the cannabinoid trials above was built around fentanyl withdrawal, and the Kentucky one used oxycodone. Fentanyl also complicates the medication start: in a survey of 1,679 people at 49 US treatment centers, taking buprenorphine within 24 hours of fentanyl carried about five times the odds of severe withdrawal.
That survey, Varshneya 2022, relied on self-report, and 22% of those who took buprenorphine within a day of fentanyl described severe withdrawal (152 of 685), against 11.6% of those who took methadone in the same window. The problem has a clinical answer, which is a prescriber who knows how to time and dose the start, and no cannabis study touches it. The FDA's own 2023 naloxone announcement described the overdose crisis as driven primarily by synthetic opioids like illicit fentanyl.
The other danger arrives after withdrawal ends. In a London follow-up of 137 people who went through 28-day inpatient detox, three died of overdose within four months of discharge, and all three were in the group that had completed both detox and the program. None of the 43 who left before finishing detox, still tolerant, died. The authors put it down to lost tolerance and the unpredictability of going back to heroin. Getting through withdrawal at home, with a gummy or without one, ends in exactly that position.
Is CBD good for opioid withdrawal or cravings?
For withdrawal, nobody has tested it. For craving, one positive trial and one flat one. Hurd 2019 gave 400 or 800mg of CBD once a day for three days to people abstinent from heroin, and craving and anxiety triggered by drug cues fell against placebo, an effect still visible a week after the last dose.
That Mount Sinai trial registered 45 participants across placebo, 400mg and 800mg arms, and it also saw lower heart rate and salivary cortisol during the cue sessions. What it measured was craving in a lab session built around drug cues, in people who were not in withdrawal. Nobody's relapse was tracked. When a 2025 pilot trial added 600mg a day of CBD to buprenorphine for 28 days in an inpatient unit, both groups saw craving and negative mood fall over time, and CBD showed no clear advantage over placebo (most of the between-group comparisons leaned toward placebo, with cue-induced craving trending the other way). A 2026 Lancet Psychiatry meta-analysis of cannabinoid trials in mental and substance use disorders found no significant effect on opioid use disorder outcomes.
At 25mg a gummy, the Hurd dose of 800mg is 32 gummies a day, and the UCLA dose is 24. The hemp aisle sells the conclusion of the first trial at about one thirtieth of its dose and skips the second trial entirely.
Can cannabis replace methadone or buprenorphine?
No study shows it can. The 2019 National Academies report concluded that methadone, buprenorphine and extended-release naltrexone are effective and save lives, and that stopping medication often leads to relapse and overdose. In 17,568 Massachusetts overdose survivors, methadone was tied to 59% lower opioid-related death and buprenorphine to 38% lower.
Those Massachusetts numbers come from Larochelle 2018 in Annals of Internal Medicine, and the uptake figures are the part that stings: in the year after a nonfatal overdose, 11% enrolled in methadone and 17% received buprenorphine. Naltrexone showed no association with mortality, though the authors said too few events occurred to draw a confident conclusion. A national claims study two years later, Wakeman 2020 in JAMA Network Open, followed 40,885 people with commercial or Medicare Advantage coverage. Only buprenorphine or methadone was linked to fewer overdoses, with an adjusted hazard ratio of 0.24 at three months and 0.41 at twelve, and only 12.5% of the cohort got either. Detox, residential stays and behavioral care alone showed no such link.
So the medication with the mortality data is the one most people never receive, and the product with none is sold at the corner store. The National Academies' wording on the stop side is the reason this page will not entertain a taper off buprenorphine or methadone in favor of cannabis: discontinuing medication, in their summary, often leads to relapse and overdose.
Does using cannabis during treatment hurt outcomes?
Mostly it shows no measurable effect either way. Lake and St Pierre's 2020 review of 41 studies across methadone, buprenorphine and naltrexone found most reported no significant association with opioid use, adherence or retention, and no naltrexone study showed worse outcomes for cannabis users.
A 2019 methadone meta-analysis in CMAJ Open, covering 23 studies, found no consensus that cannabis reduced opioid use or kept people in treatment longer, and rated the evidence very low quality. Read together, the two reviews give a clinic little reason to discharge a patient over cannabis and give the patient little reason to expect it to do the medication's work. The prescriber still needs to hear about it, and the Lake review notes that monitoring cannabis use can help a clinician shape the treatment plan. The psychiatric side of heavy THC use is on our psychosis risk page, and the parallel question for drinking on our alcohol use disorder page.
Why does every source say to carry naloxone?
Because it reverses an opioid overdose, fentanyl included, and bystanders are often there to give it. The CDC says naloxone can restore breathing within 2 to 3 minutes, that more than one dose can be needed with fentanyl, and that people with opioid use disorder should carry it and keep it at home.
The FDA approved Narcan 4mg nasal spray for over-the-counter sale on March 29, 2023, the first naloxone product cleared for use without a prescription, and the CDC now lists it as available over the counter in all 50 states. The CDC's other numbers explain the push: about 76% of the 105,007 drug overdose deaths in 2023 involved an opioid, and a potential bystander was present in nearly 43% of them. Nobody can use naloxone on themselves, which is why the CDC tells carriers to let the people around them know where it is.