Cannabis and Cancer: What the Evidence Actually Shows

Cannabis helps a lot of people get through cancer treatment. That is not the same thing as treating cancer. Almost all the confusion lives in the gap between those two sentences.

Cancer patients are using cannabis faster than physicians can tell them whether it works.

In 2024, the American Society of Clinical Oncology, the main professional body for cancer doctors in the United States, published its first full guideline on cannabis and cannabinoids in adults with cancer. The panel worked through thirteen systematic reviews and dozens of trials. It opened by conceding the obvious: patient use has run well ahead of the evidence.

Its central recommendation is that cannabis should not be used as cancer-directed treatment, meaning treatment aimed at the tumor itself, outside a clinical trial. On the one hand, there’s a real and growing body of evidence that cannabis helps people endure cancer and its treatment. On the other, a claim that has circulated for two decades with almost nothing behind it.

What the Evidence Supports

Solid ground is symptom management.

Nausea ranks highest. For patients still vomiting through chemotherapy despite standard anti-nausea drugs, ASCO says cannabinoids can be added to the regimen. This is the oldest formal recognition cannabinoids have in medicine. In 1985 the FDA approved synthetic THC, sold as dronabinol under the brand name Marinol, for chemotherapy-induced nausea. It cleared a second synthetic THC compound the same year, nabilone, sold as Cesamet. Eli Lilly pulled Cesamet from the American market in 1989 for commercial reasons. Valeant Pharmaceuticals bought the drug in 2004 and returned it to US pharmacies in 2006.

So cannabinoids have four decades in the oncology toolkit. The catch is that newer anti-nausea drugs work better and do not get patients high, which leaves cannabinoids as a second option brought in when the first ones fail.

After that, the ground softens.

Appetite is the one everybody assumes is settled, because everybody knows what the munchies are. In 2002, a trial in patients with advanced cancer tested dronabinol head-to-head against megestrol, an older appetite drug. Dronabinol lost. Patients reported wanting to eat more without putting on meaningful weight. Stimulating appetite and reversing the metabolic freefall of cancer cachexia, the wasting syndrome that comes with advanced disease, turn out to be different problems. That is one trial, now more than two decades old, in the hardest population there is. It tells you cannabinoids did not reverse late-stage wasting. It does not tell you they do nothing for appetite in someone earlier in treatment who cannot face a meal, which is the situation most patients are actually in.

Pain is where the record gets uncomfortable. Nabiximols, sold as Sativex, is a mouth spray with roughly equal parts THC and CBD, and it is the most rigorously studied cannabis medicine on the planet. It went through three Phase 3 trials in cancer pain, the stage of testing designed to prove a drug works. Marie Fallon and colleagues reported two of them in the British Journal of Pain in 2017. The third, led by Aron Lichtman, put 199 patients on nabiximols against 198 on placebo and appeared in the Journal of Pain and Symptom Management in 2018.

It missed the primary endpoint in all three.

That is a real result and it should be said plainly. It is also worth knowing what those trials were asking: whether a spray could produce a measurable additional drop in pain scores in patients with advanced cancer already on optimized opioid therapy, a setting where placebo responses run high and the room to improve is narrow. There were flickers in the secondary data. In the Lichtman trial, nabiximols beat placebo on two of three quality-of-life instruments at week three and on all three at week five, and analyses run after the fact found benefits among American patients that did not appear in patients elsewhere.

Patient experience runs ahead of that record. Plenty of people report cutting their opioid use and taking fewer painkillers, and the real-world data point the same direction. A prospective registry study of 358 cancer patients enrolled in the Quebec Cannabis Registry, published in BMJ Supportive & Palliative Care in 2023, recorded significant drops in pain severity, pain interference and total medication burden across the first nine months. It also had no control group, its authors said the findings need confirmation in randomized placebo-controlled trials, and the registry itself was supported in part by unrestricted grants from licensed cannabis producers. That is what real-world evidence looks like. Suggestive, not decisive.

Outside oncology, in chronic pain, a recent systematic review in Annals of Internal Medicine pooled 25 short-term randomized trials covering 2,303 patients, nearly two-thirds of them with neuropathic pain. Oral THC-only products moved pain about 0.78 points on a ten-point scale. Balanced THC-to-CBD oromucosal sprays, the class Sativex belongs to, moved it about half a point. Both came with moderate to large increases in dizziness, sedation and nausea. And the THC-only category split against itself: nabilone cut pain by 1.59 points, dronabinol by 0.23, which is to say not at all.

Real relief for some people. Modest on a population scale. Both things are true.

What the Lab Shows, and Why That Is Not a Cure

The question underneath all of this is whether cannabis can kill cancer.

In the laboratory, there is genuinely interesting evidence. For twenty-five years, a lab at the Complutense University of Madrid run by Manuel Guzmán has documented how cannabinoids attack tumor cells. THC can trigger a stress response inside a cancer cell that pushes it to digest and destroy itself. Cannabinoids can choke off the blood supply a tumor needs and blunt its ability to invade neighboring tissue. Glioblastoma, breast, pancreatic, melanoma. In cell cultures and in mice, the compounds do impressive work. And the work has not stopped. In June 2026, a team at Rostock University Medical Center reported that cannabigerol and cannabichromene, two non-intoxicating cannabinoids that get a fraction of the attention paid to THC and CBD, killed lung cancer cells in culture and pushed them into apoptosis, with CBG appearing to act through a receptor called PPARα rather than the cannabinoid receptors everyone assumes are doing the work. That is a cell-culture finding, with everything that implies. It is also a reminder that this field is active, not closed.

Then come the qualifiers, and they are not small ones.

Cell lines and mice are not people. Oncology’s graveyard is full of compounds that cleared tumors in a dish and did nothing in a human body. One common reason is dosage: the concentrations that kill tumor cells in vitro, meaning in glassware rather than in a patient, are often far higher than anything that can safely be put into a human bloodstream.

Cannabinoids also do not only inhibit tumors. In some studies, under some conditions, THC has stimulated cancer cell growth by suppressing the immune response that keeps tumors in check. Cancer Research UK, reviewing the same body of laboratory work, notes that different cannabinoids appear to do different things to different cancers, and that under certain conditions they can encourage cancer cells rather than kill them.

The National Cancer Institute makes the same distinction in its own summary for clinicians: antitumor activity in preclinical models, no approval and no established role in treating cancer in patients.

And when the National Academies of Sciences, Engineering, and Medicine reviewed the entire published literature in 2017, it landed on a phrase worth memorizing. There is insufficient evidence, the report concluded, either “to support or refute” the idea that cannabinoids are an effective treatment for cancers.

The Only Human Trials

In humans, the entire body of direct anti-tumor evidence fits in a few paragraphs.

Guzmán’s team ran the first one in 2006. Nine patients with terminal recurrent glioblastoma, THC delivered through a catheter directly into the tumor. It was a safety study, and it cleared that bar. The treatment did not appear to accelerate the disease. Two patients were alive a year later, though both have since died. Guzmán’s own verdict was measured: the significance, he has said, is little in clinical terms, but the trial offered “some hints towards possible anti-tumoral action of THC in certain patients.”

Guzmán has been more careful about that study than most people who cite it. In a 2013 opinion piece, he wrote that “cannabinoids are efficacious drugs to treat at least some types of cancers in laboratory animals,” but that the anecdotal evidence “remains far from supporting that cannabinoids are efficacious anticancer drugs for large patient populations.” Part of the problem is that “cancer” is not one disease. It is at least 150 histological types and thousands more by molecular profile. Nothing treats all of that, which makes the question of which therapy helps which patient enormously specific.

In 2021, a team led by Chris Twelves at the University of Leeds published the results of GWCA1208, which gave nabiximols alongside chemotherapy to patients with recurrent glioblastoma. One-year survival was 83 percent in the nabiximols group against 44 percent on placebo. Those numbers traveled fast.

What traveled less is the shape of the study. It was a Phase 1b trial built to test safety and tolerability, run in two parts. Part 1 was open-label, with six patients. Part 2 randomized 21 more, twelve to nabiximols and nine to placebo, and the survival figures come from those 21. Two of the placebo patients died within the first 40 days of enrollment. At six months, progression-free survival was identical in both arms, 33 percent. The trial was sponsored by GW Research, which made the drug, and the authors wrote that the survival difference would need testing in a randomized controlled trial.

Which is why the trial that matters is the one still running.

ARISTOCRAT is a Phase 2, triple-blind, placebo-controlled trial testing nabiximols plus temozolomide, the standard chemotherapy for the disease, against placebo plus temozolomide in patients with recurrent glioblastoma of a specific molecular subtype. It is coordinated by the Cancer Research UK Clinical Trials Unit at the University of Birmingham and funded by The Brain Tumour Charity, with the drug supplied free by Jazz Pharmaceuticals, which now owns the Sativex line. Its primary endpoint is overall survival, which is to say the only measure that answers the question people actually ask. Patients are randomized two to one, with a target enrollment of 120 across 22 UK sites, revised down from an original 234. Per the trial registry it is still recruiting, with primary completion expected in September 2026.

Where the Cure Story Came From

In 2003, a doctor removed a basal cell carcinoma from a Canadian man named Rick Simpson. It is the most common and most treatable skin cancer there is. Simpson then made a concentrated cannabis oil at home, put it on his bandages and concluded the oil had healed him. That account is his own, and it is essentially the only one there is.

What he did next is the part his critics skip. He gave the recipe away. RSO is not a branded product, and Simpson’s own site explains how to make the oil rather than selling it. He does not profit from the syringes now sitting in dispensary cases across legal markets, in rice-grain doses. Whatever else is true here, nobody was running a business.

The claim is still wrong.

There is no scientific evidence that RSO cures cancer. Not weak evidence. None. The founding observation is a lesion a surgeon had already removed, read afterward as proof that the oil had worked. That is a reasoning error, and it is one that costs other people.

The sharpest critics of the cure claim are not prohibitionists. Adam Friedman, professor of dermatology at the George Washington School of Medicine and Health Sciences, has co-authored papers on the therapeutic potential of cannabinoids and uses CBD in his own practice. He has called the phenomenon terrifying.

The cost lands on patients, not on Simpson. Some delay or refuse treatment that works. Many more use cannabis alongside treatment without telling anyone, which means nobody is watching for drug interactions, product contamination, or the possibility, suggested by early research, that cannabis may blunt the effectiveness of immunotherapy. Those findings have been disputed and need further study, but the risk is not a footnote: ASCO leaned on early evidence of poorer outcomes in patients using cannabis during immunotherapy as part of the reasoning behind its one firm recommendation, that cannabis not be used as cancer-directed treatment outside a trial.

And the silence is widespread. A cross-sectional survey of patients and survivors at the Hollings Cancer Center at the Medical University of South Carolina, in a state where cannabis is still illegal, shows a weighted prevalence of 26 percent.

Ethan Zohn, the Survivor: Africa winner and co-producer of the High Times docuseries Kicking Back, who beat CD20-positive Hodgkin’s lymphoma twice, described exactly that gap in a recent interview for this magazine. When he used cannabis to manage the side effects of chemotherapy, no doctor could tell him how to do it safely. The only guidance he got was not to smoke it.

What the Government Now Admits

For more than fifty years, the United States kept cannabis in Schedule I, the category defined in statute as having no currently accepted medical use. Heroin’s shelf. That was a political artifact rather than a scientific finding, and the government’s own machinery has now said so in writing.

In August 2023, the Department of Health and Human Services, drawing on the FDA and NIDA, formally recommended moving cannabis to Schedule III. The FDA review behind that recommendation counted more than 30,000 licensed practitioners recommending cannabis across 43 jurisdictions to more than six million patients for at least fifteen conditions, and found credible scientific support in three specific areas: pain, chemotherapy-induced nausea and vomiting, and appetite loss.

The same three symptom domains the trials keep pointing at. Not the tumor.

In April 2024, the Justice Department’s Office of Legal Counsel concluded that the DEA is bound by the scientific and medical determinations HHS reached. The agency that spent five decades insisting cannabis had no medical value was told by its own department’s lawyers otherwise.

Then, on April 23, 2026, the DOJ issued a final order moving two categories into Schedule III: cannabis contained in an FDA-approved drug product, and cannabis subject to a state medical marijuana license. It took effect five days later, the first time cannabis has moved out of Schedule I since the Controlled Substances Act became law in 1970.

It is also narrower than the headlines suggested. Everything outside those two categories, including all adult-use cannabis, remains in Schedule I. An expedited DEA hearing on broader rescheduling opened on June 29, 2026, and has not concluded. Schedule III is a meaningful change for research access and for the tax treatment of medical operators. It is not the end of federal prohibition.

Where That Leaves Patients

Here is the honest accounting.

Cannabis helps people endure cancer treatment. Nausea is the firmest case, with sleep, anxiety and the general siege of being sick showing up powerfully in patient experience even where formal trials lag behind.

Cannabinoids might eventually earn a role against certain tumors. That possibility rests on twenty-five years of laboratory work that keeps producing results and one carefully caveated survival signal in a study too small to prove anything. Unproven is not the same as disproven, and the distance between them is exactly what a trial is for. ARISTOCRAT will tell us more than everything that came before it combined.

What nobody can responsibly claim today is that cannabis treats cancer in a human body. The scientist who ran the only human trial says so himself, and says it more cautiously than most of the people who quote him. He also has not stopped working on the question, and neither has anyone else in this piece.

What is beyond dispute is that a plant with documented medical value is being used by millions of sick people who cannot get a straight answer about how to use it safely. That is not a scientific problem. It is a regulatory one.

The trial that could finally answer the question is funded in substantial part by the patients and families who wanted it answered. Neither the state nor the pharmaceutical industry would fully bankroll it. They passed the hat.

This article is for information only and is not medical advice. Cannabis can interact with cancer treatments. Patients should talk to their oncologist before starting, stopping or changing anything about their care.