What does the research say?
In a small study of 18 people with arthritis at the base of the thumb, a CBD oil was compared with a product containing only the carrier oil, and improvements in pain and hand function were reported. In a study of 60 people with jaw joint muscle pain, pain also fell clearly with the CBD product (a 70.2% vs. 9.8% reduction in pain score). In another study of 60 people with teeth grinding (bruxism) and jaw muscle pain, using a CBD gel rubbed onto the jaw muscle from inside the mouth, pain fell by 41–57%; there was no clear change in those using an inactive gel. However, these two jaw pain studies came from the same university department and their participants were mostly young people in their 20s; the first was registered after the study had ended, and the plan in the second one's registration does not match the paper either. There is no similar study from another research group (October 2026).
In contrast, in 80 patients after knee replacement surgery, CBD applied to the skin made no difference in pain, sleep or opioid use. In 40 patients with chemotherapy-related nerve pain, there was no difference between CBD cream and an inactive cream; three small studies of muscle soreness after exercise also found no difference in pain. In 40 women with persistent pain at the vulvar entrance (vestibulodynia), a gel containing CBD and a cannabis terpene (myrcene) was used for 60 days; pain fell in both groups and the difference between them was not significant, although pain during intercourse fell more with the CBD gel. A positive result was reported in people with nerve pain in the feet and legs, but this study's method was weak. In 69 people with recurrent mouth ulcers (canker sores), a 0.1% CBD paste applied to the ulcers made them shrink faster than placebo; pain, however, was lower than with placebo only on day 5, and fell on more days with a steroid paste.
These studies share common problems: the numbers of participants are small, the amount of CBD in the products differs or was not reported at all, and the carrier oil's own effect cannot be separated out. In most studies, it was also not adequately measured whether participants could tell which product they were using, or how much CBD was absorbed through the skin. So although the positive results are encouraging, it is not yet possible to draw a reliable conclusion.
There are two studies with THC-containing products rubbed on the skin. In 100 people with nerve pain caused by diabetes, a THC-dominant cannabis oil that also contained small amounts of CBD and CBN was rubbed on the skin for 12 weeks; pain symptoms fell much more than with placebo. An effect of this size has not been seen in other cannabinoid studies in nerve pain; the study was single-center, some numbers in the paper do not match each other, and there is no second study confirming the result. In 21 women with hand joint pain caused by aromatase inhibitors used in breast cancer treatment, CBD-dominant and THC-dominant ointments were compared; there was no placebo and participants knew which ointment they were using. Both ointments were well tolerated; the share of people reporting clear improvement was given as higher with the THC ointment, but the two groups were not compared statistically, and with this design the true effect of the ointments cannot be measured. The evidence based on these two studies is very weak for THC-containing products.
In a study with no control group, 18 people with hand osteoarthritis used a 4% CBD gel for 4 weeks; pain decreased and returned to its earlier level when the gel was stopped. Because there was no comparison group, it is not known how much of this improvement came from the product and how much from spontaneous improvement over time or from expectation; the company that developed the gel co-funded the study. Two small uncontrolled studies (leg pain in former athletes, localized pain in fibromyalgia) with products that contained, alongside CBD, pain-relieving substances such as camphor and methyl salicylate or other undisclosed ingredients also reported improvement; but they were not included in the table because it cannot be separated whether the improvement came from CBD. Systematic reviews evaluating studies of CBD applied to the skin are also cautious: they found a high risk of bias and drew no firm conclusion; the US Agency for Healthcare Research and Quality's 2025 review judged the evidence insufficient.
Strength of evidence: ●●○○ Weak evidence. Finding: Mixed results. What does this mean?
Ongoing research on this topic
Trials testing CBD for this topic on ClinicalTrials.gov: 3 under way (1 recruiting); results are awaited from 2 that ended in the last two years. A trial being under way does not show that CBD works.
- NCT07093138: Phase 4, recruiting, target 30 participants
- NCT07776405: Phase 2, not yet recruiting, target 50 participants
- NCT07733791: Observational study, not yet recruiting, target 60 participants
Registries in other countries also list 5 recruiting trials on this topic: other registries.
Possible side effects and drug interactions
Most studies reported no side effects or did not include them in the published abstracts; however, this does not mean the products are completely risk-free. In studies that asked about side effects in detail, the most common problem was mild skin reactions: in the study testing hand ointments, a quarter of the women had mild skin irritation; in the THC oil study, mild skin reactions such as hives and redness occurred at a rate of 10% both in those using the cannabis oil and in those using the placebo oil, meaning some of these reactions may come from the oil or from the rubbing itself. Cosmetic products may contain other active ingredients besides CBD.
In a study testing five commercial products in 46 healthy volunteers, no THC was found in the blood of those using products applied to the skin, and urine tests were not positive by US workplace criteria; however, the oral fluid test of 9 participants (7 of whom were using the same lotion) exceeded the THC threshold used in workplace tests. People who must be tested at work or in sports should know this. These studies have no data on drug interactions of CBD applied to the skin.
To check interactions with the medicines you take: CBD drug interaction checker.
How was CBD used in studies?
The products used in the studies were very different: 1 mL twice a day of an oil containing 6.2 mg of CBD per milliliter (study dose; not a personal recommendation), a 4% CBD gel three times a day (study dose; not a personal recommendation) and a cream containing about 100 mg of CBD per application (study dose; not a personal recommendation). Most studies did not report the amount of CBD in the product; so it is not possible to say what amount might work.
In the bruxism study, a 10% or 5% CBD gel was rubbed onto the jaw muscles from inside the mouth before bed (40 or 20 mg a day; study dose; not a personal recommendation); in the vulvar pain study, a 5% CBD gel was used once a day (study dose; not a personal recommendation).
Doses with THC-containing products rubbed on the skin: in the study of nerve pain caused by diabetes, 1–6 drops, depending on pain severity, of an oil containing 3.2 mg of THC per drop (the paper does not state how many times a day it was applied; study dose; not a personal recommendation); in the study of hand pain caused by aromatase inhibitors, one scoop of ointment on each hand 3 times a day (2210 mg of CBD or 375 mg of THC per jar; study dose; not a personal recommendation). In the mouth ulcer (canker sore) study, a 0.1% CBD paste was applied to the ulcer 3 times a day (study dose; not a personal recommendation).
Study details (for health professionals)
A small crossover RCT in arthritis of the thumb base joint (n=18), an RCT in jaw joint muscle pain (n=60) and an RCT with an intraoral CBD gel in jaw muscle pain due to bruxism (n=60; 10% and 5% gels better than placebo for pain, no between-group difference in teeth-grinding severity) reported positive results. No benefit was seen after total knee replacement (n=80), in chemotherapy-induced neuropathy (n=40) or in muscle soreness after exercise (three small studies). In vestibulodynia, a CBD + myrcene gel (n=40) did not differ from placebo for pain; only pain during intercourse fell more, but the baseline score was higher in this group. Samples are small and product contents vary; the carrier oil's own effect, blinding and measurement of systemic absorption are inadequate. The two positive RCTs in jaw pain come from the same university department and their participants were mostly in their 20s; the first was registered retrospectively, and the number of participants and date in the second one's registration do not match the paper; in both studies, groups were assigned by choosing numbers. In recurrent canker sores, in a 69-person, three-arm (CBD, steroid, placebo) RCT with a 0.1% CBD paste applied to the ulcers, ulcers shrank faster than with placebo, but pain was lower than with placebo only on day 5. In a study with no control group (4% CBD gel in hand osteoarthritis, n=18), pain decreased from baseline. With THC-containing products: in diabetic neuropathy, a single-center RCT with a THC-dominant oil (n=100, 12 weeks) reported a large benefit (NPSI-T difference −17.3), but it has not been confirmed and the numbers in the paper are inconsistent; in an open-label study without placebo (n=21) comparing CBD and THC ointments in hand pain caused by aromatase inhibitors, both ointments were well tolerated, and the groups were not compared statistically. The evidence for THC-containing products is very weak. Systematic reviews: high risk of bias in topical CBD studies (2023); the AHRQ 2025 living review found the evidence for topical CBD and for the THC:CBD:CBN oil insufficient; a 2025 review of medicines applied through the skin or inside the mouth in jaw disorders saw a short-term benefit in two CBD studies but did not perform a pooled analysis. In healthy volunteers (n=46), CBD passed into the blood with three of five commercial products; there was no THC in the blood, but in 9 people the oral fluid test exceeded the THC threshold. October 2026 search (PubMed, Cochrane, ClinicalTrials.gov, Europe PMC): 13 randomized studies measuring pain with cannabinoid products applied to the skin or inside the mouth (12 placebo- or vehicle-controlled; 2 with a THC-containing product) and 3 systematic reviews were found; all of them are in the table below, along with 1 study with no control group and 1 study of absorption in healthy volunteers. Cochrane has no review specific to this topic. Not included in the table: hemp seed oil products (seed oil is not a CBD or THC product), a wound pain study in which CBD was also given by mouth, two small studies with no control group using products that contained camphor, methyl salicylate or other undisclosed substances alongside CBD (it cannot be separated whether the improvement came from CBD or the other substances), publications in which pain was not measured or that are at the level of case reports, and a knee gel study published only as a conference abstract; studies whose results are unpublished are listed under the unverifiable items.
Evidence assessment: Weak; conflicting (many small RCTs, results in both directions)
CBD alone
With topical CBD, the results conflict; the positive studies are small and single-center (the two positive RCTs in jaw pain come from the same research group), and the negative studies include knee replacement (80 people) and chemotherapy neuropathy (40 people). In mouth ulcers (canker sores, 69 people), a CBD paste sped up ulcer shrinkage, but pain was lower than with placebo on only one measurement day. A small study with no control group (hand osteoarthritis, 18 people) reported improvement from baseline; because there was no comparison group, it does not count as evidence of effectiveness. Systematic reviews find the evidence insufficient or the risk of bias high (AHRQ 2025, a 2023 review of topical CBD); a 2025 review evaluating the two jaw pain studies sees a short-term benefit but makes no strong recommendation.
Products containing THC
There are two studies with THC-containing products rubbed on the skin. In nerve pain caused by diabetes, a THC-dominant cannabis oil (THC:CBD:CBN) reduced pain symptoms much more than placebo in a 100-person, 12-week, single-center placebo-controlled study (NPSI-T difference −17.3 points); some numbers in the paper do not match each other and there is no second study confirming the result. In a 21-person open-label study without placebo comparing THC-dominant and CBD-dominant ointments in hand joint pain caused by a breast cancer medicine (aromatase inhibitor), both ointments were well tolerated; the share of people reporting clear improvement was given as 50% with the THC ointment and 18% with the CBD ointment, but the two groups were not compared statistically and participants knew which ointment they were using, so these rates do not show that one ointment is better. The AHRQ 2025 review found the evidence for the THC:CBD:CBN oil insufficient.
Evidence: Very weak (a single placebo-controlled study with a THC-containing product rubbed on the skin: single-center, not confirmed by another study, some numbers in the paper inconsistent; plus a small feasibility study without placebo)
Dose range in studies
In topical studies, 6.2 mg/mL (1 mL twice a day), 4% gel 3 times a day, ~100 mg per application, intraoral 5–10% gel (20–40 mg/day), 5% CBD + myrcene gel once a day; in canker sores, 0.1% CBD paste 3 times a day; THC-dominant oil on the skin, 1–6 drops depending on pain severity (3.2 mg of THC per drop); hand ointments 3 times a day (2210 mg of CBD or 375 mg of THC per jar); most did not report the concentration
Starting dose and titration
Not reported in detail in the studies.
These are doses researchers gave under medical supervision; they are not a recommended dose for you. "mg/kg/day" means milligrams per kilogram of body weight per day.
| Study | Participants | Product | Dose and titration | Duration | Result | Source |
|---|---|---|---|---|---|---|
| Heineman 2022 Crossover, double-blind RCT | 18 Arthritis of the thumb base joint | Topical CBD 6.2 mg/mL in shea butter (control: shea butter alone) | 1 mL twice a day | 2 weeks + 1-week washout + 2 weeks crossover | During the CBD period, greater improvement in VAS pain, DASH and SANE scores than during the control period; grip and pinch strength and range of motion similar. Numbers not in the abstract (full text could not be read); each period 2 weeks, 18 people Side effects: No adverse events | PMID 35637038, doi:10.1016/j.jhsa.2022.03.002, NCT04611347, 2025, the primary measures in the registration are grip strength, PROMIS upper extremity, Kapandji and range of motion, not the pain scales highlighted in the paper) |
| Nitecka-Buchta 2019 Double-blind RCT | 60 Jaw joint (TMJ) myofascial pain | 20% cannabis extract oil in a cholesterol ointment (about 1.46% CBD; THC 0 according to the manufacturer's certificate; the extract also contains small amounts of CBG and CBC), applied over the masseter muscle | Twice a day | 14 days | The primary measure in the registration was resting sEMG activity of the masseter muscle: over 14 days, it fell by 11% on the right and 12.6% on the left with CBD, and by 0.2% and 3.3% with placebo. Pain (VAS 0–10) fell from 5.6 to 1.7 (70.2%) with CBD and from 5.1 to 4.6 (9.8%) with placebo; between-group difference on day 14 significant (p < 0.001). Participants were young (average age 23); groups were assigned by drawing an even or odd number from a container; the study was registered after it ended, and the registration lists 20 participants. From the same university department as Walczyńska-Dragon 2024; the authors reported no conflicts of interest Side effects: No adverse events recorded | PMID 31698733, doi:10.3390/jcm8111886, NCT03994640, 20 participants in the registration) |
| Walczyńska-Dragon 2024 Described as double-blind, placebo-controlled RCT, 3 arms (10% CBD / 5% CBD / placebo); according to the main publication, the CBD gels were slightly bitter and the placebo gel almost tasteless; according to the sleep publication of the same trial, the placebo gel was clear and the CBD gels slightly opaque (blinding may have been compromised) | 66 randomized; 60 completed and were analyzed (20 in each arm) Sleep bruxism (teeth clenching/grinding) and jaw joint disorder originating in the chewing muscles (Poland) | CBD gel applied to the chewing muscle from inside the mouth (prepared with 100% CBD isolate) | 40 mg/day with the 10% gel, 20 mg/day with the 5% gel (on both sides, at bedtime) | 30 days | Pain fell by 57.4% with the 10% gel and 40.8% with the 5% gel; no significant change with placebo; on day 30, both gels differed significantly from placebo in pain (VAS) and masseter muscle activity; the difference in pain between the 10% and 5% gels was not significant; no between-group difference in teeth-clenching severity (Bruxoff) (p = 0.23). Groups were assigned by choosing numbers from opaque containers, and only the 60 people who completed the study were analyzed (66 people were randomized). Participants were young (average age 24); from the same university department as Nitecka-Buchta 2019; the authors reported no conflicts of interest Side effects: No side effects reported | PMID 38592260, doi:10.3390/jcm13051417, NCT05562635, a July 2023 start and an 'orally dissolving film' title, the paper gives January–May 2023 and 66 people, PMID 40732351, PMID 41517391, which cites the same registration number, is a separate study: a pilot comparing two CBD gels in 20 students, without placebo and without measuring pain) |
| Haffar 2022 Double-blind, placebo-controlled RCT, 4 arms | 80 (19 CBD, 21 essential oil, 21 CBD + essential oil, 19 placebo) Primary total knee replacement | Topical CBD (± essential oil); concentration not reported | 3 times a day; amount not reported | 2 weeks | No difference between CBD and placebo in pain, sleep or 42-day opioid use; the only significant difference was against CBD: on day 2 after surgery, pain was higher in the CBD group than in the essential oil group (VAS 69.9 vs. 51.0; p = 0.013) Side effects: Not reported (in the abstract) | PMID 35390457, doi:10.1016/j.arth.2022.03.081 |
| Murina 2025 Multicenter, double-blind, placebo-controlled RCT | 40 (20/20) Vestibulodynia (persistent pain at the vulvar entrance) | 5% CBD + myrcene (a cannabis terpene) gel, applied to the vulvar entrance | Once a day, at bedtime (2 pumps) | 60 days | Pain (p=0.39) and the cotton swab test (p=0.06) decreased in both groups, difference not significant; pain during intercourse fell more with CBD (p<0.01), but the baseline score was higher in this group; no difference in percentage changes (full text; the abstract is written more positively) Side effects: No one dropped out; no side effects reported | PMID 41153723, doi:10.3390/biomedicines13102440 |
| D'Andre 2024 Double-blind, placebo-controlled crossover pilot RCT | 40 randomized (20 + 20); 36 people completed both periods Established chemotherapy-induced peripheral neuropathy with pain or tingling of at least 4/10 severity (US, Mayo Clinic) | Pure CBD (isolate) cream | Twice a day on the affected areas (amount of CBD in the cream not in the abstract or the registration) | 2 weeks + 2 weeks crossover | Neuropathy symptoms (EORTC QLQ-CIPN20) were similar in the CBD and placebo periods; according to the investigators, the results do not support that CBD isolate cream reduces this pain. A small, short (2 + 2 weeks) pilot study Side effects: Well tolerated; side effect scores similar in both periods; no serious side effects (registry) | PMID 39016024, doi:10.1089/can.2023.0253, NCT05388058 |
| Pastina 2025 Double-blind, placebo-controlled pilot RCT | 28 (9 CBD, 10 placebo cream, 9 no-cream controls) Muscle soreness after exercise | Topical cream, about 100 mg of CBD per application | ~100 mg per application | 3 application days | No difference in pain, pressure pain threshold or performance; on the muscle soreness questions, the group difference favored the placebo cream (the placebo cream group perceived recovery as better) Side effects: Not reported (in the abstract) | PMID 38980809, doi:10.1089/can.2024.0012, NCT06320327, the primary measure in the registration is the vertical jump test) |
| Mendes 2026 Crossover, double-blind pilot RCT | 15 Exercise-induced muscle damage | Topical gel 1% CBD (10 mg/g), 2 g per application (20 mg CBD) | 2 g of gel (20 mg CBD) | 72 hours | No difference in delayed-onset muscle soreness; only a moderate difference in isometric torque. The gels were provided by a company (Becann); the authors reported no conflicts of interest Side effects: Not reported (not in the full text either) | PMID 41877261, doi:10.1186/s42238-026-00420-0 |
| Alpy 2023 Double-blind RCT (CBD on one arm, placebo on the other arm) | 21 Exercise-induced delayed-onset muscle soreness (biceps muscle), healthy university students | Commercial CBD balm (1000 mg per container); placebo coconut oil moisturizer | Applied 30 minutes and 24, 48 and 72 hours after exercise | 72 hours | No difference between CBD and placebo on any day in arm circumference, muscle strength or pain Side effects: Not reported (in the abstract) | PMID 38287971, doi:10.70252/KAGH6991 |
| Xu 2020 Placebo-controlled RCT (followed by crossover) | 29 (15 CBD, 14 placebo) Lower limb peripheral neuropathy | Topical CBD oil (250 mg CBD / 3 fl oz); THC not reported | Not reported (in the abstract) | 4 weeks | Over four weeks, a significant reduction compared with placebo in intense pain, sharp pain, cold sensations and itch was reported; the abstract does not state blinding or the source of the product (methodologically weak; full text could not be read) Side effects: No adverse events reported | PMID 31793418, doi:10.2174/1389201020666191202111534 |
| Seevathee 2025 Double-blind, placebo-controlled, parallel phase 3 RCT (single center) | 100 randomized (50 + 50); all in the intention-to-treat analysis; 7 people did not complete follow-up Adults with type 2 diabetes and painful diabetic peripheral neuropathy in the lower limbs who were not taking neuropathic pain medicines (Thailand, Kalasin) | THC-dominant cannabis oil applied to the skin (per drop THC 3.2 mg, CBD 0.32 mg, CBN 0.65 mg); placebo MCT and coconut oil. The product was provided by the Department of Thai Traditional and Alternative Medicine of Thailand's Ministry of Public Health | 1–2, 3–4 or 5–6 drops depending on pain severity (the daily application frequency is not stated in the full text that was read) | 12 weeks | The neuropathic pain symptom score (NPSI-T total) fell from 25.6 to 5.6 with the cannabis oil and from 25.2 to 22.9 with placebo; between-group difference at week 12 −17.3 points (p < 0.001). An effect of this size has not been seen with other products applied to the skin or in other cannabinoid studies in diabetic neuropathy; the single-center result has not been confirmed by another study, and some numbers in the paper do not match each other (the difference at week 4 is −5.50 in the table and −8.14 in the text). Because the product is THC-dominant, the result cannot be attributed to CBD. 3 of the 9 authors work at the ministry department that provided the product; the authors reported no conflicts of interest Side effects: Side effects 10% in both groups, all mild (hives, redness, skin pain); resolved by the end of follow-up | PMID 39720705, doi:10.1159/000542511, Thai Clinical Trials Registry TCTR20211220008 |
| Zylla 2026 Randomized, open-label feasibility study; no placebo (CBD-dominant vs. THC-dominant ointment) | 24 registered (ClinicalTrials.gov); 21 women completed the study Women with stage 1–3 breast cancer taking an aromatase inhibitor, with drug-related joint, muscle and bone pain (AIMSS) (US, Minnesota) | Ointments provided free of charge by a medical cannabis company (Vireo Health): CBD-dominant (2210 mg of CBD per jar, THC under 0.3%) or THC-dominant (375 mg of THC per jar, CBD under 20 mg); on the hands | 3 times a day, one scoop (1.25 mL spoon) applied to the back of each hand and spread over the hand, wrist and fingers | 2 weeks randomized + 2-week open extension with the chosen ointment | The primary aim was feasibility: 71% of participants continued into the 2-week extension, and 86% of weekly questionnaires were completed. At week 2, 86% of participants reported improvement in hand symptoms (M-SACRAH); the share reporting more than 50% improvement was given as 50% in the THC ointment group and 18% in the CBD ointment group. The two groups were not compared statistically (no test in the abstract that was read); there was no placebo and participants knew which ointment they were using, so the improvements cannot be interpreted as an effect of the ointment and the rates do not show that one ointment is better. The CBD-dominant ointment also contained a small amount of THC (under 0.3%). Pain scale (BPI) results are not in the abstract that was read (full text could not be read). Supported by the US National Cancer Institute and NIH; a medical cannabis company provided the ointments free of charge; the authors reported no conflicts of interest Side effects: Mild skin irritation in 24% of participants; 1 person stopped because of the ointment's greasy texture | PMID 41467893, doi:10.1177/25785125251398286, NCT05935891 |
| Umpreecha 2023 Randomized, double-blind, placebo- and active-drug-controlled RCT, 3 arms (CBD / steroid / placebo) | 72 randomized (24 each); 69 analyzed (23 CBD, 22 steroid, 24 placebo); plus 100 and 50 healthy volunteers for safety Adults with recurrent canker sores (mouth ulcers) (Thailand, Bangkok) | 0.1% CBD oral paste applied to the ulcer (CBD from a 5% cannabis extract donated by a company; purity and THC content not reported); comparisons: 0.1% triamcinolone (steroid) paste and an inactive paste | 3 times a day, after meals, applied to the ulcer with a measuring spoon | 7 days | The main measure was ulcer size (the sample size was calculated for it): with CBD and with the steroid, ulcers shrank more than with placebo on all measurement days. Pain (daily VAS) was lower than placebo in the CBD group only on day 5 (p = 0.039); with the steroid on days 4, 5 and 7; no difference in pain between CBD and the steroid. Oral health–related quality of life improved similarly in all three groups. The same person could be re-enrolled for different canker sore episodes; registration was retrospective. Supported by Thailand Science Research and Innovation Fund; the authors reported no conflicts of interest Side effects: No allergies or side effects were seen; vital signs and blood values did not change in the volunteers in the safety phase | PMID 36803360, doi:10.1186/s12906-023-03886-0, Thai Clinical Trials Registry TCTR20220802004 (retrospective registration, August 2, 2022) |
| Bawa 2024 Open-label, single arm, feasibility | 18 (15 completed) Hand osteoarthritis | Transdermal CBD gel 4% | 0.25 mL 3 times a day (~30 mg CBD/day) | 4 weeks | About −1.9 NRS in pain (no control group); returned to baseline during the washout period. The company that developed the gel co-funded the study and donated the product; 4 of the 11 authors are employees of this company Side effects: 31 mild adverse event reports and 1 moderate; no THC detected in urine, CBD and breakdown products detected | PMID 38783008, doi:10.1038/s41598-024-62428-x, ANZCTR ACTRN12621001512819 |
| Chou 2025 (AHRQ living review) Living systematic review and meta-analysis (US Agency for Healthcare Research and Quality; search up to April 28, 2025) | Since the start, 29 RCTs (2,579 people) and 15 observational studies (49,453 people) Chronic pain in adults and adolescents (48% of studies in neuropathic pain) | Cannabinoids (grouped by THC:CBD ratio and product type) and plant-based products; CBD applied to the skin as a separate group | The evidence for CBD applied to the skin was found insufficient: no firm conclusion can be drawn. The RCT comparing a CBD gel applied inside the mouth with placebo in jaw disorder was added in the 2024 update, and the RCT testing a THC:CBD:CBN oil applied to the skin in diabetic neuropathy was added in the 2025 update; the evidence for the THC:CBD:CBN product is also insufficient. For comparison: pure CBD by mouth did not reduce pain more than placebo (4 RCTs, 334 people; moderate certainty); an oral spray containing THC and CBD in similar proportions slightly reduced pain (mean difference −0.54/10; moderate certainty). A public agency review Side effects: Dizziness, drowsiness and nausea increased with THC-containing products; no separate side effect result for products applied to the skin in the abstract | PMID 40956916, doi:10.23970/AHRQEPCCER250UPDATE2025, NCBI Bookshelf NBK618045 (AHRQ report no. 25-EHC036), PROSPERO CRD42021229579, PMID 40238954, journal publication Chou 2026, PMID 41429020 | ||
| Galant 2025 (systematic review) Systematic review of RCTs (RoB 2; no meta-analysis because the criteria for pooled analysis were not met) | 9 placebo-controlled RCTs, 390 people; 2 with CBD Jaw joint disorders, chewing muscle pain and sleep bruxism | Medicines applied through the skin or the lining of the mouth; CBD studies: cannabis extract ointment (about 1.46% CBD) and CBD isolate gel (5% and 10%) | In both CBD studies, pain fell more than with placebo over 14–30 days, and both studies were judged to have a low risk of bias; the largest percentage reduction was with the 1.46% CBD ointment. But because each medicine was tested in only one or two studies and with different methods, no pooled analysis was done; the authors write that no strong clinical recommendation can be made; no improvement in mouth opening compared with placebo. The two CBD studies come from the same university department; the review authors are independent of it, but the review does not discuss that the first study was registered retrospectively or that groups were assigned by choosing numbers in both studies. No external funding; the authors reported no conflicts of interest Side effects: No side effects were reported in the CBD studies | PMID 41463118, doi:10.3390/biomedicines13123110, PROSPERO CRD42025632021 | ||
| Scholfield 2023 (systematic review) Systematic review (no meta-analysis) | 8 studies (3 placebo-controlled RCTs, 5 single-arm); 11 more studies found in registries but not accessible Human studies with CBD applied to the skin (skin conditions or safety 2, pain 2, behavior 1; the others not detailed in the abstract) | CBD products applied to the skin (50–250 mg or 0.075–1% CBD; most combined with other substances) | The studies have a high risk of bias and incomplete reporting; because of differences in formulation and methods, no pooled conclusion could be drawn about CBD's dose, effectiveness and safety. The authors warn of the risk that users generalize poorly designed studies to all CBD products, and call for new placebo-controlled studies that measure blood CBD levels and show that the effect comes from CBD. Most of the knee replacement, chemotherapy neuropathy and exercise studies were published after this review (full text could not be read) Side effects: Side effects rare and mild; according to the authors, of limited meaning without efficacy data | PMID 35605018, doi:10.1089/can.2021.0154 | ||
| Zamarripa 2024 Randomized, double-blind, placebo-controlled pharmacokinetic and drug testing study (pain not measured) | 46 (6–8 people in each of 5 product groups, 9 on placebo) Healthy adults (US, Johns Hopkins University) | 5 commercial hemp-derived products high in CBD and low in THC: cream, lotion, patch, balm, gel; placebo | Twice a day (amounts of CBD in the products not in the abstract) | 17 days (application in the lab on the first day, 9 days of use at home twice a day, 1-week washout) | With three of the five products, CBD passed through the skin into the blood; levels rose most after 7–10 days of use, and most with the lotion containing CBD plus vitamin E, which increases absorption. No THC was detected in the blood; no urine sample was positive by US workplace testing criteria; but in 9 people (7 lotion, 1 patch, 1 gel), THC in oral fluid exceeded the 2 ng/mL "positive" threshold used in workplace tests. The products had no subjective, mental or physical effects. Funded by the US Department of Defense and SAMHSA (the substance use and mental health agency); 1 of the 15 authors works at a CBD company (full text could not be read, conflict of interest statement not seen) Side effects: Well tolerated (no details in the abstract) | PMID 38217086, doi:10.1093/jat/bkae001, NCT04741477, no results in the registry) |
Reported side effects
- Most studies reported no adverse events, or they are not in the abstract. No side effects were reported in the bruxism and vestibulodynia RCTs (in the bruxism study, 6 of 66 people did not complete the study); in the chemotherapy neuropathy RCT, side effect scores were similar in both periods and there were no serious side effects; no side effects were seen in the canker sore RCT. Skin reactions: 24% mild irritation in the hand ointment study; in the THC oil RCT, mild skin reactions at 10% in both groups (including placebo)
Who should be careful
- With topical products, urine tests were not positive, but THC ≥2 ng/mL was found in oral fluid in 9 participants (Zamarripa 2024)
- Product contents vary; cosmetic products may contain CBD together with other active ingredients
- Products rubbed on the skin can cause mild skin reactions (irritation, redness, hives); in a study testing hand ointments, mild irritation was reported in a quarter of users
Abbreviations: RCT, randomized controlled trial (a study in which participants are assigned to groups by chance); placebo, an inactive substance; SMD, OR and RR, effect measures used in analyses that pool studies; CYP, liver enzymes that break down medicines.
If you are considering CBD
- Talk with your doctor about the cause of your pain; this is especially important if the pain lasts a long time or gets worse.
- Check the product label for the amount of CBD and the other ingredients; ask for an independent laboratory certificate of analysis (COA).
- Find out whether the product contains THC; this matters if you need to be tested.
- Do not stop the medicines you use for pain on your own.
Choose your age group, medicines and situation to prepare a short summary you can take to your doctor. Nothing you enter is sent anywhere.
Bottom line
The effect of CBD applied to the skin on pain is not yet clear: although a few small studies reported encouraging results, others found no difference. The US Agency for Healthcare Research and Quality's (AHRQ) 2025 systematic review also finds the evidence for CBD applied to the skin insufficient. Until larger, well-designed studies are done, it is best to approach these products with cautious expectations.
Frequently asked questions
Does CBD oil or cream applied to the skin work for pain?
The evidence is weak and conflicting. Three small studies in arthritis at the base of the thumb and jaw muscle pain, and one methodologically weak study in nerve pain in the feet and legs, reported positive results; after knee replacement, in chemotherapy-related nerve pain, vulvar pain and muscle soreness after exercise, however, no clear difference was seen compared with an inactive product (placebo); in mouth ulcers (canker sores), pain was lower than with placebo on only one measurement day. The two positive jaw pain studies come from the same research group; the US Agency for Healthcare Research and Quality's 2025 systematic review found the evidence for CBD applied to the skin insufficient. The only placebo-controlled study with a THC-dominant cannabis oil, in nerve pain caused by diabetes, reported a large benefit, but this result has not been confirmed; the evidence for THC-containing products is very weak.
Does CBD applied to the skin get into the blood?
Some of it can, but the amount varies a lot by product. In a study testing five commercial products in 46 healthy volunteers, CBD could be measured in the blood with three products; levels rose most after 7–10 days of regular use, and most with a lotion containing CBD plus a substance that increases absorption. In the hand osteoarthritis study, CBD and its breakdown products were also found in the urine of people who used the gel for 4 weeks. In most pain studies, however, absorption was not measured; so the link between the results for pain and blood levels is not known.
Can CBD cream cause a problem on a drug test?
In a study with healthy volunteers, the urine tests of people who used five commercial products applied to the skin for 10 days were not positive by US workplace criteria, but in the oral fluid test of 9 of the 46 participants, THC was above the workplace "positive" threshold. It is important to check the label and the certificate of analysis to see whether the product contains THC.
Is CBD applied to the skin more effective than CBD taken by mouth?
No study has compared the two directly. The results of products applied to the skin conflict and the studies are small; so it is not known which is more effective.
Can CBD cream be obtained with a prescription in Turkey?
In Turkey, products derived from cannabis are regulated by the Kenevirden Elde Edilen Ürünlere Dair Yönetmelik (Regulation on Products Derived from Hemp) dated January 31, 2026. These products, including CBD, can be sold only in pharmacies. Those in the medicine and health product category are dispensed with a prescription; the regulation does not state a prescription requirement for supplement products and cosmetics with a THC content of at most 0.3%. As of October 2026, there is no approved (licensed) CBD medicine in Turkey. If you are thinking of using it, talk with your doctor first.
References
Show 18 references
- Heineman 2022. Crossover, double-blind RCT. PMID 35637038, doi:10.1016/j.jhsa.2022.03.002, NCT04611347, 2025, the primary measures in the registration are grip strength, PROMIS upper extremity, Kapandji and range of motion, not the pain scales highlighted in the paper)
- Nitecka-Buchta 2019. Double-blind RCT. PMID 31698733, doi:10.3390/jcm8111886, NCT03994640, 20 participants in the registration)
- Walczyńska-Dragon 2024. Described as double-blind, placebo-controlled RCT, 3 arms (10% CBD / 5% CBD / placebo); according to the main publication, the CBD gels were slightly bitter and the placebo gel almost tasteless; according to the sleep publication of the same trial, the placebo gel was clear and the CBD gels slightly opaque (blinding may have been compromised). PMID 38592260, doi:10.3390/jcm13051417, NCT05562635, a July 2023 start and an 'orally dissolving film' title, the paper gives January–May 2023 and 66 people, PMID 40732351, PMID 41517391, which cites the same registration number, is a separate study: a pilot comparing two CBD gels in 20 students, without placebo and without measuring pain)
- Haffar 2022. Double-blind, placebo-controlled RCT, 4 arms. PMID 35390457, doi:10.1016/j.arth.2022.03.081
- Murina 2025. Multicenter, double-blind, placebo-controlled RCT. PMID 41153723, doi:10.3390/biomedicines13102440
- D'Andre 2024. Double-blind, placebo-controlled crossover pilot RCT. PMID 39016024, doi:10.1089/can.2023.0253, NCT05388058
- Pastina 2025. Double-blind, placebo-controlled pilot RCT. PMID 38980809, doi:10.1089/can.2024.0012, NCT06320327, the primary measure in the registration is the vertical jump test)
- Mendes 2026. Crossover, double-blind pilot RCT. PMID 41877261, doi:10.1186/s42238-026-00420-0
- Alpy 2023. Double-blind RCT (CBD on one arm, placebo on the other arm). PMID 38287971, doi:10.70252/KAGH6991
- Xu 2020. Placebo-controlled RCT (followed by crossover). PMID 31793418, doi:10.2174/1389201020666191202111534
- Seevathee 2025. Double-blind, placebo-controlled, parallel phase 3 RCT (single center). PMID 39720705, doi:10.1159/000542511, Thai Clinical Trials Registry TCTR20211220008
- Zylla 2026. Randomized, open-label feasibility study; no placebo (CBD-dominant vs. THC-dominant ointment). PMID 41467893, doi:10.1177/25785125251398286, NCT05935891
- Umpreecha 2023. Randomized, double-blind, placebo- and active-drug-controlled RCT, 3 arms (CBD / steroid / placebo). PMID 36803360, doi:10.1186/s12906-023-03886-0, Thai Clinical Trials Registry TCTR20220802004 (retrospective registration, August 2, 2022)
- Bawa 2024. Open-label, single arm, feasibility. PMID 38783008, doi:10.1038/s41598-024-62428-x, ANZCTR ACTRN12621001512819
- Chou 2025 (AHRQ living review). Living systematic review and meta-analysis (US Agency for Healthcare Research and Quality; search up to April 28, 2025). PMID 40956916, doi:10.23970/AHRQEPCCER250UPDATE2025, NCBI Bookshelf NBK618045 (AHRQ report no. 25-EHC036), PROSPERO CRD42021229579, PMID 40238954, journal publication Chou 2026, PMID 41429020
- Galant 2025 (systematic review). Systematic review of RCTs (RoB 2; no meta-analysis because the criteria for pooled analysis were not met). PMID 41463118, doi:10.3390/biomedicines13123110, PROSPERO CRD42025632021
- Scholfield 2023 (systematic review). Systematic review (no meta-analysis). PMID 35605018, doi:10.1089/can.2021.0154
- Zamarripa 2024. Randomized, double-blind, placebo-controlled pharmacokinetic and drug testing study (pain not measured). PMID 38217086, doi:10.1093/jat/bkae001, NCT04741477, no results in the registry)
Points that could not be verified from the full text
- Haffar 2022 (concentration, amount, side effects), Pastina 2025 (side effects) and Xu 2020 (amount applied) were read only from the abstract
- The full text of D'Andre 2024 could not be read; the amount of CBD in the cream is not in the abstract or the registration (according to the registration, it was applied twice a day). In Alpy 2023, the amount applied is not clear in the paper
- Murina 2025: the abstract says CBD was better for pain and in the cotton swab test; in the full text these differences are not significant (p=0.39; p=0.06). The result from the full text was entered in the table
- The full text of Zylla 2026 could not be read (publisher embargo): pain scale (BPI) results and a statistical comparison between the two ointments are not in the abstract that was read. Scholfield 2023 and Zamarripa 2024 (amounts of CBD and THC in the products, conflict of interest statement) were read only from the abstract; which topical studies the AHRQ 2025 report included could not be read because the report page required bot verification
- In Seevathee 2025, some numbers do not match each other: the between-group difference at week 4 is −5.50 points in the table and −8.14 in the text; at week 8, −12.56 and −14.54; the placebo group's week 4 score is 22.22 in the table and 24.86 in the text. The daily application frequency is not stated in the paper
- A study with a synthetic CBD gel in knee osteoarthritis (Hunter 2018) was published only as a conference abstract; the abstract page could not be read because it required bot verification, and it was not included in the table
- Studies completed but with results unpublished or not posted to the registry (October 2026): NCT07162454 (CBD and urea cream in diabetic foot, 23 people, ended June 2025), NCT06771076 (CBD cream in tendon pain in athletes, 35 people, May 2025), NCT07445568 (exercise and CBD on the skin in knee osteoarthritis, 50 people, November 2025), ISRCTN29199098 (cannabis cream on the fingers in rheumatoid arthritis, pilot, 2014). NCT04103814 (CBD acid cream in big toe osteoarthritis and hallux valgus) was stopped with 11 people, and the data were not analyzed. Status unknown: NCT04585230 (CBD roll-on after knee replacement, 92 people in the registration; same center and same four-arm layout as Haffar 2022: CBD, oil, CBD + oil, placebo; very likely the same trial, but the paper's abstract has no registration number and the registration lists the oil as 'mineral oil', so the match could not be verified), NCT05908552, NCT06099171, NCT05411913. Recruiting or not yet started: NCT06108349 (carpal tunnel), NCT07638189 (knee osteoarthritis), NCT07776405 (heel pain), NCT07093138 (migraine), NCT07027566 (tooth sensitivity)
Research history
Here we record what changed on this page, when and why, with dates. The newest change is at the top.
A blinding note was added to the Walczyńska-Dragon 2024 row: the CBD gels differed from the placebo gel in taste and appearance, so participants may have noticed which group they were in (PMID 38592260 and the sleep publication from the same trial, PMID 40732351, full texts). The conclusion and the evidence level did not change.
Related studies: Walczyńska-Dragon 2024, published February 29, 2024 (PMID 38592260, doi:10.3390/jcm13051417, NCT05562635, a July 2023 start and an 'orally dissolving film' title, the paper gives January–May 2023 and 66 people, PMID 40732351, PMID 41517391, which cites the same registration number, is a separate study: a pilot comparing two CBD gels in 20 students, without placebo and without measuring pain))
The last sentence of the page's short summary was rewritten in a calmer, more guiding tone: instead of "Talk to your doctor before you start.", it now says whom you can talk with, and about what, if you are considering CBD. The conclusion and the strength of evidence did not change.
In October 2026, PubMed, Cochrane, ClinicalTrials.gov and Europe PMC were systematically searched; 7 studies were added: 2 randomized studies with THC-containing products rubbed on the skin (a placebo-controlled study with a THC-dominant oil in diabetic neuropathy and a study without placebo comparing CBD and THC ointments in hand pain caused by a breast cancer medicine), a placebo-controlled study with a CBD paste applied to mouth ulcers (canker sores), 3 systematic reviews (one of them the US Agency for Healthcare Research and Quality's living review) and an absorption study in healthy volunteers that is the source of the drug test information on the page. Two small uncontrolled studies with products that also contained other pain relievers or undisclosed substances alongside CBD were not included in the table because the improvement could not be attributed to CBD. In existing rows: the product type of the cannabis extract ointment in jaw pain, of a CBD oil rubbed on the skin and of the CBD + myrcene gel used in vulvar pain became 'CBD-rich extract, THC content unclear'; it was written that the two positive jaw pain studies come from the same research group, that the first was registered retrospectively, that the second one's registration does not match the paper, and that an article thought to be an additional publication is a separate pilot study; it was added that the only significant difference in the knee replacement study was against CBD, along with the company ties and registration numbers of the hand osteoarthritis gel study and details of the chemotherapy neuropathy study; the incorrect 'no one dropped out' statement in the side effect summary was corrected. The conclusion and strength of evidence for THC-containing products were written for the first time (very weak). The conclusion and the strength of evidence did not change.
Related studies: Seevathee 2025, published 2025 (PMID 39720705, doi:10.1159/000542511, Thai Clinical Trials Registry TCTR20211220008); Zylla 2026, published February 2026 (online December 2025) (PMID 41467893, doi:10.1177/25785125251398286, NCT05935891); Chou 2025 (AHRQ living review), published July 2025 (PMID 40956916, doi:10.23970/AHRQEPCCER250UPDATE2025, NCBI Bookshelf NBK618045 (AHRQ report no. 25-EHC036), PROSPERO CRD42021229579, PMID 40238954, journal publication Chou 2026, PMID 41429020); Galant 2025 (systematic review), published December 17, 2025 (PMID 41463118, doi:10.3390/biomedicines13123110, PROSPERO CRD42025632021); Scholfield 2023 (systematic review), published August 2023 (online May 23, 2022) (PMID 35605018, doi:10.1089/can.2021.0154); Zamarripa 2024, published March 2024 (PMID 38217086, doi:10.1093/jat/bkae001, NCT04741477, no results in the registry)); Umpreecha 2023, published February 20, 2023 (PMID 36803360, doi:10.1186/s12906-023-03886-0, Thai Clinical Trials Registry TCTR20220802004 (retrospective registration, August 2, 2022))
Older changes (4)
Display fix: internal working file names left in the text were removed; content and evidence did not change.
Newly published and previously missed clinical studies were searched in PubMed and ClinicalTrials.gov. 4 studies added: Walczyńska-Dragon 2024, Murina 2025, D'Andre 2024 and Alpy 2023. In one of the four new randomized studies, pain fell clearly (Walczyńska-Dragon 2024); in the others, no difference from placebo was seen on the main measures. With the new studies, the strength of evidence changed from "Very weak" to "Weak".
Studies that led to the change: Walczyńska-Dragon 2024, published February 29, 2024 (PMID 38592260, doi:10.3390/jcm13051417, NCT05562635, a July 2023 start and an 'orally dissolving film' title, the paper gives January–May 2023 and 66 people, PMID 40732351, PMID 41517391, which cites the same registration number, is a separate study: a pilot comparing two CBD gels in 20 students, without placebo and without measuring pain)); Murina 2025, published October 7, 2025 (PMID 41153723, doi:10.3390/biomedicines13102440); D'Andre 2024, published December 2024 (PMID 39016024, doi:10.1089/can.2023.0253, NCT05388058); Alpy 2023, published 2023 (PMID 38287971, doi:10.70252/KAGH6991)
The full texts of the studies were reread and the information on the page was checked against its sources. Corrected: side effect, product and dose information in the study table. Points that could not be verified from the full text went down from 2 items to 1. The year of one publication named on the page was corrected to the journal's print year instead of the online publication year. The conclusion and the strength of evidence did not change.
Page published. This first review examined 7 studies.