Cannabis

Cannabis belongs to Cannabaceae, a small family it shares with hops (Humulus) and, since a recent reclassification, with hackberries (Celtis) and seven other trees and shrubs once placed in a separate family (McPartland, Cannabis and Cannabinoid Research 2018). Linnaeus named Cannabis sativa in Species Plantarum in 1753, describing a fiber plant from Europe, according to the International Plant Names Index (IPNI) and McPartland. In 1785 Lamarck coined Cannabis indica for intoxicating plants from India, Southeast Asia, and South Africa, which he said differed in eight morphological characters and in their strong odor and effect on the head. Botanists have argued over the split ever since. Small and Cronquist (1976) folded indica into C. sativa as a subspecies, and McPartland supports two subspecies: subsp. sativa, the mainly European hemp type with more CBD than THC, and subsp. indica, the mainly Asian drug type with more THC than CBD. Flora of China treats C. indica as a synonym and doubts that any split is justified beyond cultivated forms. A third name, C. ruderalis, was coined by Janischewsky in 1924 for weedy Russian plants, though he himself called it “a well marked variety.” McPartland traces the genus to the northeastern Tibetan Plateau, where it diverged from hops about 27.8 million years ago, and Ren and colleagues (Science Advances 2021), sequencing 110 genomes, concluded that cannabis was first domesticated in early Neolithic East Asia.

Flora of China describes an erect annual herb 1–3 m tall, usually with male and female flowers on separate plants (dioecious) but sometimes on the same plant, with furrowed stems. The leaves are palmately compound, with 5 to 11 leaflets low on the stem and only one to three near the top; leaflets are lance-shaped to linear, typically 7–15 cm long and 0.5–1.5 cm wide, coarsely toothed, and long-pointed, dark green above and whitish green beneath with scattered brownish resin dots. Male flowers are yellowish green, nodding, and petal-less, in loose clusters up to about 25 cm long. Female flowers are green and stalkless, crowded in the upper leaf axils among glandular, leaf-like bracts, the parts used as a drug; each ripens into a flattened, finely netted achene, the “seed,” 2–5 mm long. Leaf shape varies widely. Schultes described Afghan drug plants as short (under 1.3 m) and densely branched with broad leaflets, while Lamarck’s original indica was tall and loosely branched with narrow leaflets (McPartland). The narrow-leaflet and broad-leaflet leaves in our illustration reflect this range, not reliable species or effect markers.

Cannabinoid chemistry is concentrated in glandular trichomes, tiny resin-filled hairs that Andre and colleagues (Frontiers in Plant Science 2016) describe as mostly on the female flowers. In those glands the plant joins olivetolic acid to geranyl diphosphate to make cannabigerolic acid (CBGA), the shared precursor; three enzymes then turn CBGA into THCA, CBDA, or cannabichromenic acid (CBCA). Drug-type plants are dominated by THCA and fiber hemps by CBDA, and young plants by CBCA. Ren and colleagues found loss-of-function changes in the genes for one or the other of the two competing pathways in fiber and drug cultivars, consistent with breeding that pushed plants toward high CBD or high THC. Variants with a shorter side chain, such as tetrahydrocannabivarin (THCV), are built from a related precursor. The acids lose a carbon dioxide group slowly in the plant and much faster when heated, which is why smoking, vaping, and baking activate the plant. Andre and colleagues count more than 100 terpenes, which give each strain its smell. Potency has climbed: ElSohly and colleagues (Biological Psychiatry 2016), analyzing 38,681 US seizure samples, found average THC rising from about 4% in 1995 to about 12% in 2014 while CBD fell, so the THC-to-CBD ratio rose from 14 to about 80.

Cannabis is one of the oldest cultivated plants. Ren and colleagues place its domestication in early Neolithic East Asia, and every modern hemp and drug cultivar they sequenced descended from an ancestral pool still represented by feral plants and landraces in China. Its use as a drug is also ancient: Ren and colleagues (Science Advances 2019) found chemical residues in wooden braziers from the Jirzankal Cemetery in the eastern Pamirs, dated to about 500 BCE, showing that cannabis plants with high levels of psychoactive compounds were burned in funeral rites at least 2,500 years ago. Zuardi’s history of cannabis as a medicine (Revista Brasileira de Psiquiatria 2006) traces Chinese medical use to the Pen-ts’ao ching, the oldest Chinese pharmacopoeia, traditionally linked to the legendary emperor Shen-Nung of about 2700 BCE, which recommended cannabis for rheumatic pain, constipation, female reproductive disorders, and malaria. The surgeon Hua T’o (AD 110–207) is said to have mixed it with wine as an anesthetic, though Chinese medicine came to use mainly the seeds, as a laxative. In India, the Atharva Veda names cannabis as one of five sacred plants, and Zuardi describes a long tradition of religious and medical use.

Western medicine adopted cannabis in the 19th century. In 1839 William O’Shaughnessy, a physician working in India, published “On the preparations of the Indian hemp, or gunjah,” describing its use for rheumatism, convulsions, and the muscle spasms of tetanus and rabies, and in 1845 the French psychiatrist Jacques-Joseph Moreau published a book on hashish and mental illness. Zuardi records that the Ohio State Medical Society held a conference on cannabis in 1860, that more than 100 papers on its therapeutic use followed in the second half of the century, and that extracts were sold by firms including Merck, Burroughs-Wellcome, Bristol-Myers Squibb, Parke-Davis, and Eli Lilly. Medical use declined in the early 20th century because extracts varied unpredictably in potency and because new drugs, among them chloral hydrate and barbiturates, became available. In the United States, the Marihuana Tax Act of 1937 imposed a tax of one dollar an ounce for medical use and one hundred dollars an ounce for other uses, and cannabis was removed from the US Pharmacopeia in 1941. The Controlled Substances Act placed marijuana in Schedule I in 1970 (FDA).

Science caught up later. In 1964 Gaoni and Mechoulam identified the structure of THC (Zuardi; Russo 2011). In 1990 Matsuda and colleagues cloned a brain receptor for cannabinoids, now called CB1; in 1992 Devane and colleagues isolated anandamide, a molecule from pig brain that binds that receptor; and in 1993 Munro and colleagues found a second receptor, CB2, in immune cells of the spleen, revealing the body’s own cannabinoid signaling system. FDA approved dronabinol for appetite loss in AIDS in 1992 (NCCIH); its other cannabinoid approvals are nabilone and the purified CBD medicine Epidiolex. The UK licensed Sativex in June 2010. Laws have since loosened unevenly. The 2018 Farm Bill removed hemp from the Controlled Substances Act; in 2020 the UN Commission on Narcotic Drugs removed cannabis from Schedule IV, the most restrictive list, of the 1961 Single Convention; and on April 28, 2026, a US final rule moved FDA-approved marijuana drug products and state-licensed medical marijuana to Schedule III, leaving other marijuana in Schedule I while the Drug Enforcement Administration (DEA) held a separate hearing, from June 29, 2026, on rescheduling marijuana as a whole. The fiscal 2026 agriculture appropriations law (Public Law 119-37) narrows the federal definition of hemp to count THCA toward the 0.3% limit and to exclude finished products with more than 0.4 mg total THC per container. The Congressional Research Service (CRS) reports that the change was scheduled to take effect on November 12, 2026, and that Congress has debated delaying most of it.

Cannabis works mainly because it borrows a signaling system the body already has. Matsuda and colleagues (Nature 1990) cloned the first cannabinoid receptor, CB1, a G protein-coupled receptor found in the brain regions known to bind cannabinoids and more responsive to psychoactive than non-psychoactive cannabinoids; they suggested it underlies the changes in mood and thinking people feel. Devane and colleagues (Science 1992) then found the body’s own key for that lock, anandamide, a fatty molecule derived from arachidonic acid. Munro and colleagues (Nature 1993) identified CB2, expressed not in the brain but in immune cells. The Sativex product information explains that these receptors sit mostly at nerve endings, where endocannabinoids act as a feedback brake on the release of neurotransmitters such as glutamate. Each plant cannabinoid touches this system differently, which is why THC, CBD, CBN, and their relatives cannot be treated as interchangeable. Pertwee’s review (British Journal of Pharmacology 2008) is a useful map: THC partly switches on both receptors, CBD can block other compounds from switching them on, and THCV behaves as a blocker at low doses and an activator at higher ones.

THC (delta-9-tetrahydrocannabinol) is, in Grotenhermen’s words (Clinical Pharmacokinetics 2003), the main source of the plant’s effects, both the high and the medicinal benefits. As a partial agonist at CB1 and CB2 receptors, it mimics the effects of the body’s endocannabinoids (Sativex product information). Route changes everything. Inhaled, its effects begin within seconds to minutes, peak after 15 to 30 minutes, and fade within two to three hours; swallowed, they start 30 to 90 minutes later, peak at two to three hours, and last four to twelve hours, which is why taking a second edible because the first seems not to work can lead to an unexpectedly strong effect. Swallowed THC passes through the liver, which converts part of it into 11-hydroxy-THC, its first metabolite, and then into THC-COOH, the main breakdown product in urine (Sativex product information). Stored in body fat, THC and CBD can linger for up to four weeks. Delta-8 THC is a close chemical cousin that FDA says occurs only in tiny amounts in the plant; concentrated delta-8 is typically made from hemp-derived CBD using chemicals that FDA calls potentially harmful, and FDA warns that delta-8 has psychoactive and intoxicating effects similar to delta-9 THC.

CBD (cannabidiol) does not cause a high and works differently. Epidiolex’s US prescribing information states that its anticonvulsant mechanism is unknown and that it does not appear to work through cannabinoid receptors. In an abuse-potential study described in the same label, single doses of 750 to 4,500 mg of CBD produced drug-liking scores in the placebo range, unlike dronabinol, a synthetic THC. CBD is not inert, though: it is a moderate inhibitor of the liver enzyme CYP2C19 and affects several other drug-handling enzymes, the source of its interactions, and its long half-life of about 56 to 61 hours means it builds up with daily dosing. Taken with a high-fat meal, CBD’s peak blood level rose about fivefold and total exposure fourfold. A popular idea holds that CBD “balances” THC. Englund and colleagues (Neuropsychopharmacology 2023) tested this in 46 healthy volunteers who inhaled 10 mg of THC with 0, 10, 20, or 30 mg of CBD; THC impaired memory and caused short-lived psychotic symptoms, and no dose of CBD lessened either effect, leading the authors to conclude that the CBD-to-THC ratios common in medical and recreational products do not protect against THC’s acute harms.

The minor cannabinoids are now sold on their own, often ahead of the evidence. CBN (cannabinol) is not one of the three cannabinoids the plant’s enzymes make from CBGA; Lavender and colleagues (Journal of Sleep Research 2026) describe it as an oxidation by-product of THC. In rats, Arnold and colleagues (Neuropsychopharmacology 2025) found that CBN altered sleep patterns and that its metabolite 11-hydroxy-CBN activates CB1 receptors much as THC does. CBG (cannabigerol) is the neutral form of CBGA, the shared precursor of THC and CBD, and Russo notes that conventional breeding has produced plants rich in CBG, THCV, or CBC for study. THCV (tetrahydrocannabivarin) is THC’s shorter-chained relative; in Jadoon and colleagues’ diabetes trial (Diabetes Care 2016) it was described as non-psychoactive at 5 mg twice daily, but a closely related compound, delta-8-THCV, produced euphoric mood at the highest doses tested and made 78 of 79 urine drug screens positive for THC in an 18-person safety study (Peters and colleagues, Cannabis and Cannabinoid Research 2023). CBC (cannabichromene) is the third product of CBGA and the dominant cannabinoid in young plants (Andre and colleagues). Acid forms such as THCA and CBDA count too: the 2025 federal hemp definition adds THCA to THC, because heat turns one into the other.

Terpenes are cannabis’s other big chemical family. Andre and colleagues note that more than 100 terpenes have been identified, that they give each variety its smell and taste, and that they probably influenced which plants people selected. Russo (British Journal of Pharmacology 2011) proposed an “entourage effect,” in which terpenes such as myrcene, limonene, α-pinene, linalool, and β-caryophyllene, along with minor cannabinoids, shape THC’s effects; he presented it as a hypothesis worth testing rather than an established fact. Some evidence fits the idea that labels follow aroma rather than chemistry: Watts and colleagues (Nature Plants 2021) genotyped more than 100 commercial samples and found that those labeled “Sativa” and “Indica” were genetically indistinct across the genome, with labels tracking only a few terpenes controlled by a small set of genes. McPartland concluded that ubiquitous interbreeding has made the vernacular terms “almost meaningless,” despite the widespread claim that “Sativa” energizes and “Indica” sedates. In practice, the dose of THC, the amount of CBD, the route, and the person matter far more than a strain name, and the Englund trial is a reminder that plausible synergy does not always appear in people.

The National Academies’ 2017 report, which reviewed more than 10,000 study abstracts and reached nearly 100 conclusions, remains the broadest assessment. It found conclusive evidence that oral cannabinoids prevent and treat chemotherapy-induced nausea and vomiting, substantial evidence that cannabis or cannabinoids are effective for chronic pain in adults, and substantial evidence that oral cannabinoids improve patient-reported MS spasticity. For most other conditions the evidence was limited, insufficient, or absent. The committee also noted that the plant’s Schedule I status impeded research. FDA’s approvals follow the same contours: purified CBD for rare epilepsies, and dronabinol and nabilone for chemotherapy nausea or AIDS-related appetite loss. These approvals cover specific products at specific doses, and FDA has not approved cannabis itself for any condition. Most of the shop products people buy have never been tested in trials like those described below, and results from a pharmaceutical extract at a controlled dose cannot be assumed for a dispensary flower, a gummy, or a hemp CBD oil, whose contents, NCCIH warns, may not match their labels.

Epilepsy is CBD’s clearest success. Devinsky and colleagues (New England Journal of Medicine 2017) randomized 120 children and young adults with Dravet syndrome to 20 mg/kg/day of CBD or placebo on top of their usual drugs; convulsive seizures fell from a median of 12.4 to 5.9 a month on CBD, against 14.9 to 14.1 on placebo, and 5% became seizure-free compared with none on placebo. In Lennox-Gastaut syndrome, Thiele and colleagues (Lancet 2018) found that drop seizures fell by 43.9% on CBD and 21.8% on placebo in 171 patients, and Devinsky and colleagues (New England Journal of Medicine 2018) found reductions of 41.9% at 20 mg/kg and 37.2% at 10 mg/kg against 17.2% on placebo in 225 patients. Thiele and colleagues (JAMA Neurology 2021) found that 25 mg/kg cut seizures by 48.6% against 26.5% on placebo in 224 people with tuberous sclerosis complex. The drug’s developer, GW Pharmaceuticals, funded the trials. NICE recommends CBD, with clobazam, for Lennox-Gastaut and Dravet syndromes, and CBD for tuberous sclerosis complex, stopping if seizure frequency has not fallen by at least 30%. These doses, often hundreds of milligrams a day, dwarf those in shop CBD oils.

MS spasticity is the main use of Sativex, a spray of two cannabis extracts giving 2.7 mg of THC and 2.5 mg of CBD per puff. According to its UK product information, the first phase 3 trial found a statistically significant but clinically questionable difference of 0.5 to 0.6 points on a 0–10 spasticity scale, and a second 14-week trial found no significant effect. A third trial used a different design: Novotna and colleagues (European Journal of Neurology 2011) gave 572 people with resistant MS spasticity four weeks of Sativex, then randomized the 241 eligible responders to continue or switch to placebo for 12 weeks; those who continued held their improvement, and the groups differed by 0.84 points, with fewer spasms and less sleep disruption on Sativex. This is why NICE advises a four-week trial, continuing only if spasticity falls by at least 20%. NCCIH also cites a 2014 American Academy of Neurology guideline and a 2010 analysis of 666 people in which about a third of those taking nabiximols improved by 30% or more. A separate trial of Sativex in 72 children with cerebral palsy or brain injury was negative.

Chronic pain is the most common reason people give for medical cannabis, and the evidence has grown less encouraging over time. NCCIH describes a 2018 review of 47 studies with 4,743 participants in which 29% of people taking cannabinoids reached a 30% reduction in pain against 26% on placebo, a small difference. The 2026 Cochrane review of chronic neuropathic (nerve) pain by Ateş and colleagues, covering 21 studies with 2,187 participants, found no clear evidence that THC-dominant products achieved substantial (50%) pain relief, while nervous-system side effects rose; balanced THC-CBD products increased the proportion with moderate (30%) relief by 7 percentage points, which the authors did not consider clinically relevant, and there was no evidence for CBD-dominant products. In two phase 3 trials in advanced cancer pain not controlled by opioids, Sativex missed its primary goal (Fallon and colleagues, British Journal of Pain 2017). NICE advises that nabilone, dronabinol, THC, and THC with CBD should not be offered to manage chronic pain, and that CBD should be used for chronic pain only in a clinical trial.

Nausea, appetite, and several other uses have older or smaller studies. For chemotherapy nausea, NCCIH cites a 2015 review of 23 studies with 1,326 participants, mostly from the 1980s and 1990s; NICE suggests considering nabilone for adults whose chemotherapy nausea persists despite optimized conventional treatment. Dronabinol was approved for appetite loss with weight loss in AIDS in 1992, based on a study of 139 people (NCCIH). Glaucoma is a cautionary tale: cannabis lowers eye pressure only briefly, so NCCIH says it is not a useful treatment. A Cochrane review found no evidence that cannabis brings inflammatory bowel disease into remission. For Tourette syndrome there are two small studies with 36 people in total, and for post-traumatic stress disorder (PTSD) a nabilone study of 10 people with nightmares (NCCIH). THCV lowered fasting blood glucose more than placebo in Jadoon and colleagues’ 13-week pilot trial of 62 people with type 2 diabetes, though its primary target, HDL cholesterol, was unchanged. None of these findings establishes cannabis as a treatment.

Sleep and anxiety are the fastest-growing consumer uses, especially for CBD, CBN, and CBG products. NCCIH says the evidence on cannabis and sleep is limited and uncertain. For CBN, Corroon (Cannabis and Cannabinoid Research 2021) concluded that evidence was insufficient to support sleep claims. In the largest trial since, Bonn-Miller and colleagues (Experimental and Clinical Psychopharmacology 2024) randomized 293 adults with sleep complaints; 20 mg of CBN reduced night awakenings and overall sleep disturbance, but the main measure, sleep quality, did not differ significantly from placebo, and adding CBD did not help. Lavender and colleagues (2026) gave 20 adults with insomnia single 30 or 300 mg doses of CBN; neither changed time awake after falling asleep, the main outcome, though 300 mg improved time to fall asleep and perceived quality. In 63 veterans, CBG made no difference to sleep (Emerson and colleagues 2026). For anxiety, Bergamaschi and colleagues (Neuropsychopharmacology 2011) found that 600 mg of CBD reduced anxiety during a simulated public speech in 24 people with social anxiety, and Cuttler and colleagues (Scientific Reports 2024) found that 20 mg of CBG reduced anxiety and stress without impairment in 34 healthy adults. These are single-dose or short studies.

Cannabis reaches people in very different forms. Dried female flowers are smoked or vaporized; concentrates and oils are vaped, dabbed, or dropped under the tongue; and edibles, drinks, and capsules are swallowed. The route sets the timing: inhaled THC acts within minutes and fades in two to three hours, while swallowed THC takes 30 to 90 minutes to start and lasts four to twelve hours (Grotenhermen). The dose depends on potency, which has risen steeply; ElSohly and colleagues found average US seizure samples at about 12% THC by 2014, three times the 1995 level, and concentrates are far stronger. Hemp-derived CBD oils, gummies, and creams, and products containing CBN, CBG, or delta-8 THC, are widely sold, but NCCIH warns that cannabis and CBD products may be contaminated or mislabeled, and FDA notes that delta-8 products have not been evaluated or approved for safe use. Hemp seed foods are a separate category: FDA has no questions about hulled hemp seed, hemp seed protein, and hemp seed oil as safe food ingredients, noting that the seeds pick up only trace THC and CBD from other plant parts and cannot make people “high.”

Licensed medicines have exact doses. Epidiolex is an oral solution of 100 mg of CBD per milliliter; for Lennox-Gastaut and Dravet syndromes the label starts at 5 mg/kg/day, divided into two doses, rising to a maintenance dose of 10 mg/kg/day and up to 20 mg/kg/day, and for tuberous sclerosis complex it targets 25 mg/kg/day. Because food changes absorption, doses should be taken consistently with or without food, and the drug should be stopped gradually. Sativex is sprayed onto the lining of the mouth, varying the site each time, starting with one spray in the evening and rising by about one spray a day to a maximum of 12, at least 15 minutes apart; the median dose in trials was eight sprays a day, and a meal can more than double the amount absorbed. Consumer CBD sits at the other end of the scale. The UK Food Standards Agency (FSA) advises healthy adults to limit CBD from food to 10 mg a day, about four to five drops of a 5% oil, and the European Food Safety Authority (EFSA) set a provisional safe intake in 2026 of about 2 mg a day for a 70 kg adult for highly purified CBD supplements, against Epidiolex doses that can exceed 1,000 mg a day.

Practical points, for adults in places where cannabis is legal: assume any product affects driving and do not drive or operate machinery after using it. With edibles, wait at least two hours before considering more, since effects can take that long to peak, and remember that they may last most of the day. Prefer products from regulated sources that state their THC and CBD content per serving, and be wary of very high-potency concentrates; the Di Forti study linked daily use of cannabis with 10% THC or more to the highest psychosis risk. Avoid illicit THC vape cartridges, which NCCIH links to serious lung injury, and treat delta-8 products as intoxicating. Store every cannabis product, especially sweets and baked goods, locked up and out of children’s sight, in its original labeled packaging. Do not combine cannabis with alcohol. If you take prescription medicines, ask a pharmacist before using CBD or cannabis, and tell your doctor, even where it is legal, because the interactions below are real. Never stop a prescribed epilepsy medicine to switch to a shop CBD oil.