Medical Cannabis for Chronic Pain: Benefits, Limitations and Treatment Decisions

Medical cannabis collection · Expanded 11 October 2026

Some cannabinoid medicines provide small average improvements in chronic pain, particularly in studies of neuropathic pain. The benefit is not universal, and adverse effects can limit treatment. “Cannabis works for pain” is therefore too broad a claim. The useful questions are which pain condition, which preparation, what size of benefit and at what cost to alertness and daily function.

Evidence checked: 1 October 2026. This article is educational and does not replace a pain assessment or an individual prescribing decision.

“small improvements in pain severity”

AHRQ, 2025 review — finding for comparable THC:CBD oral spray

The review also identifies adverse effects and limitations in the evidence. This finding is specific to a product category and a mostly short-term evidence base.

Chronic pain is not one diagnosis

Neuropathic pain follows damage or disease affecting the somatosensory nervous system. Pain associated with arthritis, injury or inflammation has different mechanisms. Chronic primary pain can itself be the main clinical problem, without another condition adequately explaining its impact. Several mechanisms can coexist in one person.

This matters because evidence from one group cannot automatically be applied to another. A trial in people with nerve pain does not establish effectiveness for every form of back pain, migraine, fibromyalgia or cancer pain. A good assessment records the diagnosis, previous investigations, current function and what treatment is supposed to change. NICE’s chronic pain guideline makes this distinction between chronic primary and secondary pain.

What the evidence suggests

The AHRQ living review finds small short-term pain improvements with some THC-containing preparations, alongside increased dizziness, sedation and nausea. Much of the evidence concerns neuropathic pain and studies lasting one to six months. Evidence for whole-plant products, longer-term outcomes and comparisons with other active treatments is more limited.

In its 2024 synthesis, comparable-ratio THC/CBD oral sprays improved pain by approximately half a point on a ten-point scale compared with placebo. That average can contain both people who benefit and people who do not. It should neither be sold as dramatic relief nor used to dismiss an individual’s worthwhile improvement. The question is whether the benefit for that person is large enough to justify continuing.

The formulation matters. A result for a standardised mouth spray cannot validate every oil, edible or dried flower sold under the cannabis label. Nor can a brief study establish that an initially helpful medicine remains effective and tolerable after years.

What about CBD without THC?

CBD is often presented as a pain treatment that avoids intoxication. The absence of a THC high is attractive, but effectiveness still needs to be demonstrated. In an eight-week randomised trial of 86 patients with knee osteoarthritis, oral CBD added to paracetamol did not provide significantly greater pain relief than placebo added to paracetamol. This was a specific preparation and regimen; it is not a verdict on every possible formulation, but it directly challenges broad analgesic claims.

Animal experiments showing changes in inflammatory signalling are not equivalent to clinical evidence of meaningful pain relief. A patient needs to know whether a treatment improves the experience of pain and their ability to live with it, not only whether it changes a biological marker.

Why people can report larger improvements than trials

Patient experience and controlled trials answer different questions. Someone starting a new treatment may also receive more clinical attention, change other medicines or begin treatment during an unusually difficult period. Pain fluctuates. People who find a product helpful may be more likely to continue treatment and complete follow-up questionnaires.

These possibilities do not invalidate a person’s experience. They explain why uncontrolled reports cannot reliably measure the drug’s independent contribution. Randomisation and a comparison group help estimate that contribution, although trials also have limitations, including short follow-up and difficulty keeping participants unaware of an intoxicating treatment.

The UK guidance needs to be stated accurately

NICE NG144 recommends against offering several THC-containing medicines for chronic pain in adults and limits CBD for this purpose to clinical trials. That NHS guidance is not equivalent to saying that every private specialist prescription is unlawful. Prescribing legality, product licensing, evidence of benefit and NHS funding are separate questions.

A private assessment should still explain the evidence, alternatives, uncertainties and follow-up arrangements. Paying for a consultation or meeting a clinic’s screening criteria does not prove that treatment will help. Equally, restricted NHS access does not mean that a patient’s pain is unimportant.

Measure function as well as pain intensity

A useful treatment goal is concrete. Examples include being able to prepare a meal, tolerate a short walk, work for a planned period or sleep with fewer pain-related awakenings. These are illustrative goals to agree with a clinician, not universal targets.

A pain score alone may miss the trade-off between relief and impairment. Someone may report less pain but become too dizzy to move safely. Another person may have only a modest score change yet regain an activity they value. A structured review should record pain, function, sleep, adverse effects and the person’s own priorities.

• Establish a baseline before starting or changing treatment.
• Choose a small number of outcomes that matter in everyday life.
• Record other treatment changes that could explain improvement.
• Agree when to review and what would count as insufficient benefit.
• Include a clinician-led plan for reducing or stopping an ineffective medicine.

Adverse effects and other pain medicines

THC-containing products can cause sleepiness, dizziness, altered judgement and anxiety. These effects may be particularly consequential when combined with other medicines that impair alertness. NCCIH’s chronic pain overview places cannabinoid benefits alongside these limitations and the uncertainty about long-term use.

Do not abruptly reduce opioids, gabapentinoids or other prescribed medicines because cannabis has been started. Whether a change is appropriate depends on the drug, the condition and a supervised plan. Evidence that cannabis reliably reduces opioid requirements remains insufficient in the AHRQ review. “Opioid-sparing” should be a demonstrated outcome, not a promise used to justify treatment.

A place within broader pain care

Medication is only one part of a pain-management plan. Rehabilitation, movement adapted to ability, psychological support and treatment of relevant underlying disease may all be important. These approaches are not statements that pain is imaginary. They address the many ways persistent pain affects sleep, activity, confidence and participation.

The strongest decision is an explicit one: identify the intended benefit, acknowledge the uncertainty, check harms and revisit the result. Continuing a medicine because it is labelled “medical cannabis” is less useful than establishing whether the particular treatment improves this person’s life.

The 2025 update: what the evidence covers

The 2025 AHRQ executive summary brings the review to 29 randomised trials and 15 observational studies. Its searches ran to 28 April 2025. That is the evidence-search date, which is different from the date this web article was expanded.

The review groups preparations by THC:CBD ratio and formulation. This prevents a result for a measured oral spray from quietly becoming a claim about every flower or oil. It also leaves several important questions unresolved, including whether benefits persist, how cognitive effects influence daily life, and whether opioid use reliably falls. An individual review should therefore record function and adverse effects alongside the pain score.