Walk into any dispensary, pharmacy, or wellness shop and the CBD products are pitched at arthritis before anything else. Creams for the knees, tinctures for the hands, capsules for the inflammation. It is the single largest consumer application of CBD, and arthritis is the condition where the distance between what the cannabinoid market claims and what the controlled human evidence supports is at its widest. Two well-run randomized trials have now tested oral CBD for arthritis pain in humans. Both came back null. One of them also found more liver enzyme elevations in the CBD group than in placebo.
The animal data, to be clear, is real. Hammell 2016 applied transdermal CBD to rats with induced arthritis and measured reduced joint swelling, reduced immune cell infiltration, and less pain behaviour, dose-dependently, with no evident side effects. Malfait 2000 found oral CBD worked in a mouse model of collagen-induced arthritis. Verrico 2020 ran a proper randomized, double-blind, placebo-controlled trial of daily CBD in dogs with osteoarthritis and found it improved their pain. The mechanism has human plausibility behind it too: Richardson 2008 confirmed that CB1 and CB2 receptors are present in the synovial tissue and fluid of both osteoarthritis and rheumatoid arthritis patients, so the target exists in the right place. None of that is fabricated. The problem is what happened when the same idea was tested in people.
Vela 2022 gave 136 patients with hand osteoarthritis or psoriatic arthritis 20 to 30mg of CBD daily for 12 weeks. No separation from placebo on pain, sleep quality, anxiety, or depression. The natural objection was that the dose was too small, and Pramhas 2023 in the Lancet Regional Health Europe went after exactly that: high-dose oral CBD added to paracetamol in 86 patients with painful knee osteoarthritis, eight weeks, a well-designed trial from the Medical University of Vienna. WOMAC pain scores dropped 2.5 points on CBD and 2.4 points on placebo, p equals 0.80. Adverse events were more frequent on CBD than placebo. Liver aminotransferase and GGT elevations above baseline showed up in 15 CBD patients against 5 on placebo. The high-dose escape hatch closed.
What the research actually shows about cannabis for arthritis
Sorting the arthritis evidence means separating three layers that get blurred together in marketing copy: preclinical animal work, human trials of CBD alone, and human trials of THC-containing products. They say different things.
The preclinical layer is uniformly encouraging and it is where the consumer story originates. Rats, mice, and dogs given CBD show less joint inflammation and less pain behaviour across multiple models and multiple research groups. The Verrico canine trial is worth singling out because dogs with spontaneous osteoarthritis are a better model of the human disease than induced arthritis in rodents, and it was a real randomized controlled trial. CBD helped those dogs. That result is sound.
The human CBD-alone layer is where it falls apart. Vela 2022 and Pramhas 2023 are the two published randomized controlled trials of oral CBD for arthritis pain, and neither found benefit over placebo. They used different doses, different joints, different populations, and different countries, and they agree with each other. The AHRQ living systematic review on cannabis for chronic pain, now in its final update, has tracked this literature continuously and finds the low-THC-to-CBD product category consistently weak. The newer CANOA trial of a CBD-rich cannabis oil in knee osteoarthritis, published in 2025, keeps the question open by testing a full-spectrum oil instead of isolate, and the field is watching whether whole-plant preparations behave differently from purified CBD. What cannot be claimed in 2026 is that oral CBD isolate has been shown to relieve arthritis pain in humans, because the trials that asked found the opposite.
The THC-containing layer is small and it is the only positive one. Blake 2006 in Rheumatology tested Sativex, a roughly balanced THC and CBD oromucosal spray, against placebo in 58 rheumatoid arthritis patients over five weeks. The Sativex group showed statistically significant improvements in pain on movement, pain at rest, quality of sleep, and disease activity score. It was a small, preliminary study, and no larger trial has replicated it in the twenty years since, which is its own commentary on how cannabis research gets funded. It remains the only randomized controlled trial in arthritis to show a positive result, and it contained THC.
Read the three layers together and the consumer narrative inverts. The market sells CBD for arthritis on the strength of animal studies. The human trials of CBD alone are null with a liver signal attached. The one human trial that worked used a product with THC in it. A reader who came to this page planning to buy a CBD tincture for their knees is being sold the cannabinoid with the weaker human case.
Osteoarthritis and rheumatoid arthritis are different diseases
Lumping them under one word is a habit of consumer marketing, and it hides a distinction that changes the entire risk calculation.
Osteoarthritis is mechanical and degenerative. Cartilage wears down, bone changes, the joint hurts and stiffens, and the process is driven by load, age, injury history, and genetics. Inflammation is present but secondary. Critically, medicine has no disease-modifying drug for osteoarthritis. Treatment is symptom control (paracetamol, NSAIDs, injections), exercise and weight management to reduce load, and eventually joint replacement. Because nothing on the shelf alters the disease course, a patient adding cannabis for pain is not displacing a treatment that would have protected their joints. The trade-off is about symptom relief and side effects, and that is a reasonable trade-off to weigh.
Rheumatoid arthritis is autoimmune and inflammatory. The immune system attacks the synovium, and untreated, the process erodes cartilage and bone and deforms joints permanently. Here medicine does have disease-modifying drugs, and they work: methotrexate, leflunomide, sulfasalazine, and the biologic agents that target TNF, IL-6, and B cells. Early aggressive DMARD treatment is why the severe joint deformity that characterized rheumatoid arthritis a generation ago is far less common now. This is the reason the DMARD boundary is the firmest clinical line on this page. Cannabis relieving RA pain while the autoimmune process continues unopposed is not a neutral outcome, because the joint damage accrues silently and does not reverse.
Psoriatic arthritis, which the Vela trial included alongside hand osteoarthritis, sits closer to the rheumatoid side: inflammatory, systemic, and managed with DMARDs and biologics. The same boundary applies.
The protocol, and why it is not CBD-leaning
Most cannabis-for-arthritis guidance recommends a CBD-dominant approach on anti-inflammatory logic. The human trial evidence does not support that, so this page recommends something different: a balanced starting protocol of 2.5 to 5mg THC with 5 to 10mg CBD, taken in the evening, titrated slowly across two to three weeks. The ratio tracks Blake 2006, the one arthritis trial that produced a positive result, and it tracks the broader chronic pain evidence where the THC component consistently carries the analgesic work with CBD modulating the side-effect profile.
Evening dosing suits the condition. Arthritis pain and stiffness interfere with sleep, and morning stiffness is a defining feature of inflammatory arthritis in particular. A dose taken a couple of hours before bed addresses the pain, the sleep disruption it causes, and does its work during the window when THC's cognitive effects matter least. Patients whose pain is worst during the day can add a small morning dose once the evening dose is established and tolerated.
Keep the CBD dose modest, and this is where the Pramhas liver finding earns its place in the protocol. The trials that produced transaminase elevations used high-dose CBD. At 5 to 10mg alongside THC, CBD is playing a supporting role and staying far below the doses where the liver signal appeared. Someone taking 100mg or more of CBD daily for arthritis, which the market actively encourages, is at the dose range where the trials found both no benefit and a measurable liver effect. That combination is the worst of both directions.
For hand and knee arthritis, topical products come up constantly, and the honest answer is that a transdermal CBD gel trial in knee osteoarthritis missed its primary endpoint while hitting some secondary ones, which is a weak result presented optimistically. Edibles work systemically and reach the synovial tissue through the bloodstream, and with the human topical evidence no stronger than the human oral evidence, the format choice comes down to preference more than proven advantage.
Interactions: the liver comes first
Arthritis patients are among the most medicated populations in chronic disease, and the interaction picture reflects it.
Methotrexate is the first concern and the most specific. It is the anchor DMARD in rheumatoid arthritis, it is hepatotoxic, and patients on it get scheduled liver function tests for exactly that reason. High-dose CBD produced significant transaminase and GGT elevations in the Pramhas trial. Adding that to methotrexate, or to leflunomide which carries its own hepatic risk, without the rheumatologist knowing means one of the drugs stressing the liver is invisible to the person monitoring it. If you take a DMARD and you want to try cannabis, that conversation happens first.
Paracetamol and NSAIDs come next. Paracetamol is hepatically metabolized and it is the background analgesic most arthritis patients are already taking, which was the reason the Pramhas trial tested CBD as an add-on to it. NSAIDs bring their own gastrointestinal and cardiovascular ceilings on long-term use. The cumulative hepatic load across paracetamol, a DMARD, and high-dose CBD is the kind of thing that shows up in bloodwork before it shows up in symptoms.
The broader cytochrome P450 story applies here as across the Pain cluster: CBD inhibits CYP3A4, CYP2C9, and CYP2D6, slowing the clearance of drugs that run through those enzymes and raising their blood levels. Arthritis patients on warfarin, certain antidepressants for comorbid pain or mood, or a proton pump inhibitor alongside their NSAID are all in that territory. The doses recommended on this page sit well below where these effects become pronounced, and the point stands regardless: the prescriber managing the polypharmacy should know about the cannabis.