Is Marijuana Medicine? What the Trials Actually Show
Medical marijuana is legal in 41 states, and almost none of it has been through a clinical trial. Real cannabinoid medicines exist: the FDA has approved four of them, and SAM supports every one. What state programs sell is different: raw flower and concentrates at recreational potency, recommended for pain, PTSD and anxiety that the trials say they do not treat, by a card mill that never sees the patient again. Here is what the FDA approved, what the big reviews found, who actually holds the cards, and what the government’s own witness admitted under oath in 2026.
Why Colorado’s 52,617 medical cardholders say they qualify
Share of active patients reporting each condition, March 2026 (patients may list more than one)
Cancer, seizures, glaucoma and HIV/AIDS, the conditions the campaigns were run on, together account for about 6 percent of cardholders.
Colorado Department of Public Health and Environment, Medical Marijuana Registry, March 2026
Four cannabinoid medicines went through trials. The plant never has.
“To date, the FDA has not approved a marketing application for cannabis for the treatment of any disease or condition.” It has approved these.
| Medicine | What it is | Approved for | Approved |
|---|---|---|---|
| Epidiolex | Purified, plant-derived cannabidiol (CBD), oral solution | Seizures in Lennox-Gastaut syndrome and Dravet syndrome, ages 2 and up; later tuberous sclerosis complex | June 2018 |
| Marinol | Dronabinol, synthetic THC, capsule | Chemotherapy nausea; AIDS-related weight loss | May 1985 |
| Cesamet | Nabilone, a synthetic THC analog, capsule | Chemotherapy nausea | December 1985 |
| Syndros | Dronabinol, oral solution | Chemotherapy nausea; AIDS-related weight loss | July 2016 |
What approval means
A known molecule, a fixed dose, a trial against placebo, a label listing the risks and a pharmacist who fills it. The FDA’s words: “Without this review, the FDA cannot determine whether a drug product is safe and effective. It also cannot ensure that a drug product meets appropriate quality standards.” A 20 percent THC flower sold under a “medical” card has none of those things.
Schedule I never blocked it
Epidiolex was approved while marijuana sat in Schedule I. So was GHB’s prescription form. The path to marijuana medicine has always run through trials, not through rescheduling the raw plant so dispensaries can call gummies medicine. What Schedule III would and would not change.
The largest evidence reviews of the decade reached the same verdict
“Evidence from randomized clinical trials does not support the use of cannabis or cannabinoids for most conditions for which it is promoted, such as acute pain and insomnia.”
Hsu et al., JAMA, 2025, review of more than 2,500 articles, December 2025
The same review: evidence-based guidelines “do not recommend the use of inhaled or high-potency cannabis (≥10% or 10 mg Δ9-THC) for medical purposes,” and 29 percent of medical users met criteria for cannabis use disorder. It found support for the approved uses: chemotherapy nausea, HIV weight loss and certain pediatric seizures.
“Given the scarcity of evidence, the routine use of cannabinoids for the treatment of mental disorders and SUDs is currently rarely justified.”
Wilson et al., The Lancet Psychiatry, 2026, 54 trials, 2,477 participants, March 2026
“There were no significant effects on outcomes associated with anxiety, anorexia nervosa, psychotic disorders, post-traumatic stress disorder, and opioid use disorder,” and “an absence of RCT evidence for the treatment of depression.” Adverse events were 1.75 times more likely on cannabinoids. The authors’ own warning: “The long-term use of these medicines could result in harm and delay the use of more effective treatments.”
Wilson et al., The Lancet Psychiatry, 2026; University of Sydney, March 2026
Chronic pain: the honest picture
This is the claim with the most behind it, and the one dispensaries lean on. The National Academies in 2017 “found evidence to support that patients who were treated with cannabis or cannabinoids were more likely to experience a significant reduction in pain symptoms,” and a 2023 BMJ umbrella review found a 30 percent pain reduction across conditions, “but increased psychological distress.” Then the higher-quality trials came in. The 2026 Cochrane review of 21 trials found “no clear evidence” that THC-dominant medicines relieve neuropathic pain by half, with more nervous-system side effects; the 2018 version put the number needed to treat at 20 and the number needed to harm at 3. The International Association for the Study of Pain, after two and a half years of appraisal, “does not currently endorse general use of cannabis and cannabinoids for pain relief.”
National Academies, 2017; Solmi et al., BMJ, 2023; Cochrane, 2026; Cochrane, 2018; IASP position statement, 2021
What the BMJ umbrella review of 101 meta-analyses rated “high” certainty
“Cannabis based medicines, cannabis, and cannabinoids resulted in poor tolerability across various conditions.” The same review found evidence of benefit for MS spasticity, chemotherapy nausea and epilepsy, the conditions the approved medicines already cover.
What people are told it treats, and what the trials found
| Condition | Trial evidence | What the research says | Source |
|---|---|---|---|
| Chemotherapy nausea | Supported | Conclusive evidence for oral cannabinoids. That is why dronabinol and nabilone are FDA-approved and prescribed today. | NASEM 2017 |
| Severe childhood epilepsy | Supported | Purified CBD (Epidiolex) reduces seizures in Lennox-Gastaut, Dravet and tuberous sclerosis. Not the same as dispensary CBD or THC. | FDA |
| MS spasticity | Supported (oral) | Nabiximols, a pharmaceutical oral spray, improved spasticity with moderate certainty. Smoked flower was not what was studied. | Cochrane 2022 |
| Chronic and neuropathic pain | Mixed, weakening | Early reviews found modest benefit; the 2026 Cochrane update found no clear evidence of meaningful relief from THC medicines, with more side effects. IASP does not endorse use. Surgical patients who use marijuana reported more pain and needed 58 percent more opioids. | Cochrane 2026; IASP; ASA 2020; ASA 2022 |
| PTSD | Not supported | The first placebo-controlled trial of smoked marijuana for PTSD in veterans found no difference from placebo. The 2026 Lancet Psychiatry meta-analysis found no effect. PTSD is the no. 2 qualifying condition in Oregon and no. 3 in Colorado. | Bonn-Miller 2021; Wilson 2026 |
| Anxiety and depression | Not supported | No significant effect on anxiety and no randomized trial evidence at all for depression. In a randomized trial, getting a medical card produced no improvement in anxiety or depression and nearly tripled the odds of cannabis use disorder. | Wilson 2026; Gilman 2022 |
| Insomnia | Weak, conflicting | Self-rated sleep improved in some trials; objective sleep got worse. In a 30-year cohort, recent and cumulative use were associated with poorer measured sleep quality. | Gilman 2022; Vu 2026 |
| Glaucoma | Not supported | Marijuana lowers eye pressure for three to four hours, then it wears off, and it lowers blood pressure to the optic nerve. The American Academy of Ophthalmology does not recommend it. The government’s own expert at the 2026 hearing: the risks “substantially outweigh the modest and short-lived intraocular pressure reduction.” | AAO, via Ophthalmology Times; Hearing transcript, day 3 |
| Opioid substitution | Reversed | A 2014 study found 25 percent fewer opioid deaths in medical marijuana states. Extended to 2017, the association flipped to 23 percent more. A 2024 analysis through 2020 found no association either way. In a four-year cohort of pain patients on opioids, marijuana users had more pain and no opioid-sparing effect. | Bachhuber 2014; Shover 2019; Nguyen 2024; Campbell 2018 |
| Nausea in general | Backfires with heavy use | Heavy use causes cannabinoid hyperemesis syndrome, a cycle of severe vomiting. Emergency visits for it rose 13-fold in Ontario from 2014 to 2021 and more than doubled in a Kaiser California cohort. | Myran 2022; Costales 2024 |
“Supported” means randomized trials of a specific, dosed product found benefit. In every supported case, that product is an FDA-approved medicine or a pharmaceutical preparation, not dispensary flower or concentrate.
State “medical” programs are recreational programs with a doctor’s note
Oregon medical cardholders after recreational stores opened in 2015
When the same product became available without a card, 84 percent of “patients” stopped renewing. Colorado’s registry fell by half, from 105,695 to 52,617, over the same decade. Oregon’s own report notes that seven providers account for more than a third of all applications, and “there are currently no medical-only cannabis dispensaries or processing sites in Oregon.”
Oregon Health Authority, OMMP statistical snapshot, July 2026; CDPHE, April 2016; CDPHE, March 2026
Same product, different label
Across 8,505 products in 653 dispensaries, medical and recreational potency were nearly identical. The authors: these concentrations “seem unsuitable for medicinal purposes, particularly for patients with chronic neuropathic pain,” where the evidence is for products under 5 to 10 percent THC. How potency got here.
The card itself is a risk factor
In a randomized trial of 186 adults in Boston, getting a medical marijuana card right away “led to a higher incidence and severity of CUD; resulted in no significant improvement in pain, anxiety, or depressive symptoms” (adjusted odds of cannabis use disorder 2.88). Among midlife adults, medical users were more likely than recreational users to report cannabis use disorder symptoms, 33.8 versus 22.5 percent, because they use more often; among near-daily users, half had symptoms. Is marijuana addictive?
Gilman et al., JAMA Network Open, 2022; Terry-McElrath and Patrick, Drug and Alcohol Dependence, 2026; Dawson et al., 2024
Medical first, recreational next
Eleven states that legalized recreational marijuana by ballot had passed a medical measure first, some by more than a decade. Medical laws themselves raised use: states that passed them saw illicit use and cannabis use disorder rise more than states that did not. Most adults in “medical” programs are not medical-only users; a national survey found about 10 percent of adult users use solely for medical purposes and 36 percent for both. Marijuana laws by state.
Ballotpedia; Hasin et al., JAMA Psychiatry, 2017; Hasin et al., JAMA Psychiatry, 2023; NASEM, 2017
“Would not have passed.”
For decades, “currently accepted medical use” under federal law meant passing a five-part test: known and reproducible chemistry, adequate safety studies, adequate and well-controlled studies of efficacy, acceptance by qualified experts and widely available scientific evidence. In 2023 HHS replaced it with a two-part test built around how many state programs allow a condition. At the DEA’s rescheduling hearing in June 2026, SAM’s lawyers asked the government’s FDA witness, Dr. Dominic Chiapperino, whether marijuana would have met the old standard. Over the DEA’s objection, he answered: “Would not have passed.”
Under the new test, FDA ran its scientific analysis on seven conditions and found support for three: pain, anorexia associated with a medical condition, and chemotherapy nausea. PTSD, anxiety, epilepsy and inflammatory bowel disease were tested and did not pass. PTSD is nonetheless a qualifying condition in 38 states.
Marijuana Moment, June 30, 2026; Marijuana Herald, hearing transcripts, August 2026; National Law Review; Sabet, STAT, April 2026
“A key official from the DEA admitted marijuana has failed the test for currently accepted medical use.”
Kevin A. Sabet, SAM, closing the hearing, July 16, 2026
The DEA’s own scientist, subpoenaed by SAM, testified that her section’s December 2024 review concluded marijuana failed the five-part test because it lacks standardized chemistry and dosing, sufficient well-controlled studies and expert consensus. The full rescheduling story.
Medicine through the FDA, not through a card mill
Research and approval. SAM supports cannabinoid medicines that pass trials, backed the 2022 federal law expanding marijuana research, and wants the components of the plant studied the way any drug is. Patients deserve a dose, a label and a pharmacist.
Where state programs exist, make them medical. Cap THC at the levels the pain evidence actually covers, require a real physician relationship instead of a telehealth card mill, ban products designed for recreation, and audit providers who issue hundreds of recommendations a year.
No new conditions by lobbying. PTSD, anxiety and depression were added to state lists by campaigns, not trials, and the 2026 evidence says marijuana does not treat them and may make them worse. States should remove conditions the trials have failed.
Tell the truth to patients. Dispensaries should be barred from making medical claims, and a product sold under a “medical” card should carry the same warnings about psychosis, addiction, heart risk and pregnancy as any other. The rest of the industry’s claims, checked.
Medical marijuana: common questions
Is marijuana FDA-approved medicine?
The plant is not. The FDA states that it has not approved a marketing application for cannabis for the treatment of any disease or condition. It has approved four cannabinoid medicines that went through trials: Epidiolex (purified CBD) for severe childhood seizure disorders, and dronabinol (Marinol, Syndros) and nabilone (Cesamet) for chemotherapy nausea and AIDS-related weight loss. SAM supports all of them.
Does medical marijuana work for chronic pain?
The evidence is weak and getting weaker. Early reviews, including the National Academies in 2017, found modest benefit. The 2026 Cochrane review of 21 trials found no clear evidence that THC-based medicines relieve neuropathic pain by half, with more nervous-system side effects, and the International Association for the Study of Pain does not endorse general use of cannabis for pain. Surgical patients who use marijuana report more pain and need more opioids.
Does marijuana help PTSD, anxiety or depression?
The trials say no. The first placebo-controlled trial of smoked marijuana for PTSD in veterans found no difference from placebo, and a 2026 meta-analysis of 54 trials in The Lancet Psychiatry found no significant effect on anxiety or PTSD and no randomized evidence at all for depression. In a randomized trial, getting a medical card produced no improvement in anxiety or depression and nearly tripled the odds of cannabis use disorder.
Can you get addicted to medical marijuana?
Yes. Pooled across studies, 29 percent of people using marijuana for medical reasons met criteria for cannabis use disorder, comparable to recreational users. Among midlife adults, medical users were more likely than recreational users to report symptoms of cannabis use disorder, 33.8 versus 22.5 percent, because they use more often.
Does medical marijuana reduce opioid deaths?
No. A 2014 study found 25 percent fewer opioid deaths in medical marijuana states, but when the same analysis was extended through 2017 the association reversed to 23 percent more deaths, and a 2024 study through 2020 found no association either way. In a four-year cohort of pain patients on opioids, marijuana users had more pain and no reduction in opioid use.
What is the difference between medical and recreational marijuana?
In the store, almost nothing. Across 8,505 products in 653 dispensaries, medical products averaged 19.2 percent THC and recreational products 21.5 percent, and 70 to 91 percent of both were above 15 percent THC. When Oregon and Colorado opened recreational stores, 84 percent and 50 percent of medical cardholders stopped renewing.
Related: Marijuana in Schedule III · Marijuana myths and facts · Marijuana and mental health · Is marijuana addictive?
Patients deserve medicine, not marketing.
SAM has model language to turn state programs back into medical programs: potency limits, real physician oversight and an end to card mills.
