Pain is the reason most people seek medical cannabis. It is also the indication where the gap is widest between how confident the conversation sounds and how uncertain the evidence actually is. A Cochrane review published in January 2026 has made that gap harder to ignore.
Do not stop or reduce a prescribed analgesic because of this article. Nothing here is a reason to change any medication. Abruptly stopping opioids or other prescribed pain medicines can be dangerous. In Thailand, cannabis flower is a controlled herb dispensed only against a prescription on form PT 33 from an authorised practitioner. Every change belongs with the person who prescribed it.
1. Not All Pain Is the Same Problem
Before reading any study on this subject, the most useful thing to understand is that “pain” is not one condition, and cannabinoids do not act on the different kinds in the same way.
| Type | What it is | Typical examples |
|---|---|---|
| Nociceptive | Normal signalling from tissue damage | A fracture, a burn, post-surgical pain |
| Neuropathic | Damage or disease affecting the sensory nerves themselves | Diabetic neuropathy, post-herpetic neuralgia, phantom limb pain, trigeminal neuralgia, HIV-associated neuropathy |
| Nociplastic | Altered central pain processing without identifiable tissue or nerve damage | Fibromyalgia and related conditions |
Chronicity is a separate axis. Acute pain is a signal. Chronic pain persists beyond healing and is better understood as a condition of the pain system itself.
This matters because the clinical evidence for cannabinoids is concentrated almost entirely in chronic neuropathic pain. Studies of acute pain, such as post-operative pain, have generally not shown benefit. The likely explanation involves the different roles endocannabinoids play in the two, but the mechanism behind the difference is not fully understood. A patient told “cannabis helps with pain” has been told something too general to act on.
2. The Mechanism
The endocannabinoid system is deeply involved in how the body modulates pain, which is why the hypothesis was worth testing in the first place.
Endocannabinoids are not stored and released like classical neurotransmitters. They are produced on demand under conditions of stress or injury. Tissue injury and inflammation trigger a rapid local rise in endocannabinoid levels, which modulates signalling in immune and other cells.
Two receptor populations matter. CB1 receptors are dense in the regions of the brain and spinal cord that gate pain signalling, and they are also present on peripheral nerve endings. Activating them dampens transmission at several points along the pathway. CB2 receptors sit mainly on immune cells and are implicated in the neuroinflammation that helps maintain chronic pain states.
The mechanism is real and well characterised. What follows is a lesson this forum returns to often: a good mechanism does not guarantee a clinically useful effect. The size of the effect is a separate question, and only trials can answer it.
3. The Trial That Gets Cited Most
The study most often invoked in this discussion is the McGill trial: Ware and colleagues, Canadian Medical Association Journal, 2010 [1]. It deserves to be described accurately rather than gestured at.
Design. A randomised, double-blind, placebo-controlled crossover trial with four 14-day treatment periods. Twenty-three adults enrolled and 21 completed. Mean age was 45.4 years. All had post-traumatic or post-surgical neuropathic pain lasting at least three months and were on stable analgesic regimens.
Intervention. Four cannabis potencies: 0% (placebo), 2.5%, 6% and 9.4% THC. Each was inhaled as a single 25 mg dose through a titanium pipe, three times daily for five days per cycle, followed by a nine-day washout.
Result. On an 11-point pain intensity scale, the 9.4% THC preparation produced a mean score of 5.4, against 6.1 for placebo. That is a difference of 0.7 points, 95% CI 0.02 to 1.4.
Secondary outcomes. Sleep improved significantly at 9.4% THC: participants fell asleep more easily and faster, with fewer periods of wakefulness. The anxiety/depression item of the EQ-5D quality-of-life questionnaire also improved compared with placebo. However, the trial found no difference in mood on its mood scale or in overall quality of life.
Adverse effects. The most common drug-related adverse effects at 9.4% THC were headache, dry eyes, a burning sensation in the painful areas, dizziness, numbness and cough. The paper reports 248 mild and six moderate adverse events across the trial, and no serious ones.
Now look at the number. The difference is 0.7 points on an eleven-point scale, and the lower bound of its confidence interval is 0.02, which only just excludes zero. The result is statistically significant, and it is small. Thresholds for clinically meaningful pain reduction are usually set considerably higher.
That is not a criticism of the study. It was careful, honestly reported and influential for good reason. The criticism is of how it gets quoted. “A randomised controlled trial showed smoked cannabis reduces neuropathic pain” is accurate. Without the effect size, it is also misleading.
A related crossover trial at the University of California, San Diego (Ellis and colleagues, Neuropsychopharmacology, 2009 [2]) tested smoked cannabis in HIV-associated neuropathic pain. Cannabis was added to participants’ existing analgesic regimens, and it reduced pain more than placebo. Note the framing: added to, not instead of.
4. What the Evidence Base Looks Like in Aggregate
Individual trials in this field are small. The way to read a field of small trials is to pool them, and that has now been done again with current methods.
Ateş, Häuser and colleagues, Cochrane Database of Systematic Reviews, January 2026 [3] updates the standing Cochrane review on cannabis-based medicines for chronic neuropathic pain in adults. It covers 21 trials and more than 2,100 adults, with treatment periods of two to 26 weeks.
Its findings:
- There is no high-quality evidence that cannabis-based medicines (THC-dominant, CBD-dominant or balanced THC/CBD) relieve neuropathic pain better than placebo.
- Balanced THC/CBD products showed small improvements, but these fell short of clinical significance.
- The quality of most of the trials is too poor to draw firm conclusions.
- THC-containing products were associated with more dizziness and drowsiness and possibly more withdrawals from trials because of side effects. Adverse events were reported inconsistently, so certainty even about this is low to very low.
- The authors call for larger studies lasting at least 12 weeks that include people with the physical and mental health comorbidities real patients have.
This is the current position of the most methodologically conservative body in evidence synthesis, published in January 2026. It is more negative than most public discussion assumes.
Two earlier guidance documents complete the picture. In March 2021, the International Association for the Study of Pain (IASP) stated that it does not endorse the general use of cannabinoids for pain, citing the lack of high-quality evidence [4]. Later in 2021, a BMJ Rapid Recommendation [5] made a weak recommendation: clinicians may offer a trial of non-inhaled medical cannabis or cannabinoids to adults with chronic pain when standard care is not enough. The panel judged the benefits small and closely balanced against the harms, which makes it a case for shared decision-making. Neither document claims cannabis is an effective first-line pain treatment.
One medicine has a separate regulatory history. Nabiximols (Sativex), a THC/CBD mouth spray, is licensed in a number of countries, mainly for MS-related spasticity. In 2005, Canada also granted it conditional approval as an add-on treatment for neuropathic pain in multiple sclerosis. See bai99 and the MS article for detail.
5. The Placebo Problem, Which Is Larger Here Than Almost Anywhere
Pain is a subjective endpoint, and placebo responses are strongest on subjective endpoints.
Gedin and colleagues, JAMA Network Open, 2022 [6] pooled 20 double-blind, placebo-controlled trials of cannabis-based therapies in adults with clinical pain, totalling 1,459 participants. Placebo alone produced a significant, moderate-to-large reduction in pain intensity (Hedges g = 0.64), and trials with low risk of bias showed larger placebo responses. In the main results, the paper reports a large effect for the active cannabinoid (g = 0.95). The difference between active drug and placebo was not statistically significant (g = 0.32, P = .09).
The same analysis found heavy media coverage of these trials, with a strong positive bias. Neither the number of positive news items nor attention scores were associated with how large the effects actually were.
This does not mean the pain relief patients experience is imaginary. Placebo analgesia is a genuine physiological phenomenon, and the relief is real to the person feeling it. The implication is narrower and harder: it is very difficult to establish how much of that relief comes from the cannabinoid, and the trials done so far have largely failed to establish it.
6. What a Modest Average Effect Means for One Person
A small average difference in a trial is compatible with several different underlying realities, and the distinction matters enormously to an individual patient.
It could mean everyone gets a small benefit. It could also mean that a minority get a substantial benefit and most get none. This responder pattern averages out to something modest. Trials that measure means cannot easily tell these apart, and research has not yet identified reliable predictors of who responds.
This is the honest reason a supervised individual trial can be reasonable even where the average effect is unimpressive. The question for a patient is not “what is the mean effect across 2,100 people” but “am I one of the people it helps?” The only way to find out is a structured trial with a defined endpoint and a defined stopping point.
Two things make that legitimate rather than wishful. It has to be supervised, so that the answer is judged rather than hoped for. And it has to have a stopping rule agreed in advance, so that “it isn’t working yet” does not become an indefinite escalation.
Some patient surveys add a finding worth putting alongside the trial data (see bai165). Some respondents say conventional analgesics relieved their pain more effectively, but they chose cannabis anyway because they preferred its side-effect profile. That is a coherent, rational preference, and it is a different claim from “it works better.” Any honest account of this field has to hold both.
7. Two Things This Article Is Not Saying
It is not saying cannabis does nothing for pain. The mechanism is real, some trials are positive, sleep improvement is reported consistently, and the patient preference described above is legitimate.
It is not a basis for stopping anything. Whether cannabis lets patients reduce their opioid doses is a separate, contested question, covered in bai181. Whatever the answer, abruptly stopping a prescribed opioid or other analgesic is dangerous. That decision belongs to a prescriber, not a forum.
8. In Thailand
Cannabis flower is a controlled herb under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568, in force since 26 June 2025 [7]. Supply requires a prescription on form PT 33 from an authorised practitioner for their own patient. Each prescription may cover no more than 30 days, and the flower must come from sources certified to the Department of Thai Traditional and Alternative Medicine’s good agricultural and collection practice (GACP) standard. Online sale and advertising of flower are not permitted. A separate Cannabis Control Act is still only a draft.
The Thai Department of Medical Services (DMS) guidance on medical cannabis lists neuropathic pain among the conditions for which there is research evidence of benefit. The same guidance says cannabis products should not be the first-line treatment: standard treatment comes first. That guidance predates the 2026 Cochrane update, which reads the trial evidence more cautiously.
If you use cannabis for chronic pain under a prescription: what kind of pain is it, and what actually changed? Was it the pain intensity itself, your sleep, or how much the pain bothered you? Those are three different outcomes, and the trials suggest they may not move together. Please describe outcomes rather than doses, and do not advise anyone to change their medication.
References
- Ware MA, et al. Smoked cannabis for chronic neuropathic pain: a randomized controlled trial. CMAJ. 2010;182(14):E694–E701. doi:10.1503/cmaj.091414
- Ellis RJ, Toperoff W, Vaida F, et al. Smoked medicinal cannabis for neuropathic pain in HIV: a randomized, crossover clinical trial. Neuropsychopharmacology. 2009;34:672–680. doi:10.1038/npp.2008.120
- Ateş G, Welsch P, Klose P, Phillips T, Lambers B, Häuser W, Radbruch L. Cannabis-based medicines for chronic neuropathic pain in adults. Cochrane Database of Systematic Reviews. 2026, Issue 1. doi:10.1002/14651858.CD012182.pub3
- International Association for the Study of Pain. IASP Position Statement on the Use of Cannabinoids to Treat Pain. 18 March 2021. https://www.iasp-pain.org/advocacy/iasp-statements/iasp-position-statement-on-the-use-of-cannabinoids-to-treat-pain/
- Busse JW, et al. Medical cannabis or cannabinoids for chronic pain: a clinical practice guideline. BMJ. 2021;374:n2040. doi:10.1136/bmj.n2040
- Gedin F, Blomé S, Pontén M, et al. Placebo response and media attention in randomized clinical trials assessing cannabis-based therapies for pain: a systematic review and meta-analysis. JAMA Network Open. 2022. doi:10.1001/jamanetworkopen.2022.43848
- Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568 (Royal Gazette, 25 June 2025). Summary: https://med-cannabis.dtam.moph.go.th/law/2418/
This article is part of the Asiannabis Community educational series on medical cannabis. It is general information, not medical advice. Do not change any prescribed treatment on the basis of anything written here.