24 +DC
States: adult-use legal
40 + DC allow medical cannabis. Federal law is separate (see below).
3
FDA-approved THC drugs
All synthetic THC (Marinol, Syndros, Cesamet). None is the plant. Epidiolex is CBD, not THC.
~4.8×
Odds of psychosis
Daily high-potency use vs. never-use, in susceptible people (EU-GEI, 2019).
~9%
Of ever-users get dependent
Lowest of the major substances — but ~1 in 6 if use starts in the teens.
How Much Evidence Actually Exists
PubMed
THC is unusual: there's a huge lab/animal literature and a large but mostly small-and-short human-trial literature — yet very few large, definitive U.S. trials, because Schedule I status throttled rigorous American research for decades. The gap at the top is a policy artifact, not a lack of interest.
What the Strongest Human Trials Show
PubMed
Where THC has real, measured benefit — the number is the finding, not the hope. These are the lanes where the evidence is strongest.
Chemo nausea & vomiting
OR 3.82
Higher odds of a complete antiemetic response vs. placebo. The basis for FDA-approved dronabinol & nabilone.
→ Whiting, JAMA 2015 · PMID 26103030
MS spasticity (nabiximols)
p = 0.0002
Pivotal Phase 3 RCT: significant drop in spasticity score vs. placebo (241 randomized). Approved in many countries as Sativex.
→ Novotna, Eur J Neurol 2011 · PMID 21362108
Chronic / neuropathic pain
NNTB 11–20
Real but small: 21% got ≥50% pain relief vs. 17% on placebo. Cochrane: benefits "might be outweighed by potential harms."
→ Mücke, Cochrane 2018 · PMID 29513392
Appetite / AIDS wasting
+38%
Appetite improvement on dronabinol vs. +8% placebo (n=139); weight stabilized. Short-duration data only.
→ Beal, J Pain Symptom Manage 1995 · PMID 7730690
The Evidence Table — Benefits AND Harms
PubMed
The honest ledger: the benefit studies above, plus the harm studies that get talked about less. Both are real.
| # | Study | Type | n | What it found |
|---|---|---|---|---|
| 1 | Cannabinoids for medical use (benchmark review) |
Meta-analysis 79 RCTs | 6,462 | Moderate / mixed |
| 2 | Cannabis for chronic neuropathic pain |
Meta-analysis 16 RCTs | 1,750 | Small benefit |
| 3 | Nabiximols for MS spasticity (pivotal Phase 3) |
RCT | 241 | Positive |
| 4 | Dronabinol for AIDS-related anorexia |
RCT | 139 | Positive |
| 5 | High-potency cannabis & first-episode psychosis (EU-GEI, 11 sites) |
Case-control | 2,138 | Harm signal |
| 6 | Transition from use to dependence (NESARC) |
Cohort | 7,389 | Harm signal |
| 7 | Cannabis use & cardiovascular events (BRFSS) |
Cross-sectional | 434,104 | Harm signal |
| 8 | Cannabinoid Hyperemesis Syndrome (CHS) |
Review + cases | — | Harm signal |
Reality check: THC's approved, evidence-backed uses are narrow — chemo nausea, MS spasticity, AIDS appetite. The broad "cannabis cures everything" claims run ahead of the data. And the harms are real but dose-, potency-, and age-dependent — heaviest for daily high-potency use, teenagers, and people prone to psychosis. Not mythical; not catastrophic for the average occasional adult.
Cost vs. the Proven Alternatives
Market data
In every therapeutic lane THC competes in, there's an established generic that costs a fraction as much — and is usually covered by insurance, while cannabis usually is not. Approx. U.S. prices, ~30-day supply.
Dosing Reality
Clinical / NIH
There is no single "safe dose." Effects rise with dose, potency, and how often it's used — and the same dose hits a teenager and an adult very differently.
Consumer start
2.5 mg THC
"Start low, go slow" — wait up to ~2 hrs for edibles before more
Research unit
5 mg THC
NIH's "standard THC unit" — for study comparability, not a recommended serving
Rx — appetite
2.5 mg ×2/day
Dronabinol (Marinol) starting dose, before lunch & dinner
Rx — chemo nausea
5 mg/m²
Dronabinol before chemo, repeat q2–4h; max 15 mg/m²/dose
Critical caveat
Potency & age matter most
High-potency product + daily use + a developing (teen) brain is where harm concentrates
The Legal Reality (2026)
T1 · Regulatory
2026 was a turning point — but a partial one. Federal law finally split the difference between "medicine" and "drug," while most recreational cannabis stayed Schedule I.
DEA / Federal
Partly rescheduled, April 2026
A final order moved FDA-approved cannabis products and state-licensed medical marijuana to Schedule III. Recreational, bulk, and synthetic THC remain Schedule I. A broader rescheduling hearing was set to begin June 29, 2026.
FDA
3 approved THC drugs — all synthetic
Dronabinol (Marinol, Syndros) and nabilone (Cesamet) are approved for chemo nausea and AIDS-related appetite loss. The plant itself is not FDA-approved. Epidiolex is CBD, not THC.
WADA / USADA
Banned in-competition only
A violation only if urinary carboxy-THC tops 150 ng/mL (raised from 15 in 2013 so out-of-competition use isn't penalized). Not banned out-of-competition.
→ USADA
What the Community Reports
Anecdotal
This section is anecdotal. These are community and patient reports — not controlled, not blinded, not weighed as evidence. They're included because the community's experience often runs ahead of (or alongside) the formal research, in both directions.
Pain & cancer patients
Pain · nausea · appetite
Widely report help with all three, and some reduce opioid use — a consistent theme that runs ahead of the modest RCT effect sizes.
Veteran / PTSD communities
Sleep & anxiety
Frequently report subjective relief — but controlled trials stay weak/mixed, and higher THC doses can worsen anxiety (the biphasic effect).
→ Sharpe 2020 · PMID 33008420
Heavy daily-user forums
"Only hot showers help"
A rising signal of CHS (cyclic vomiting) and difficulty quitting — the community corroborating the dependence and hyperemesis literature.
→ Schreck 2017 · PMID 29132050
The Bottom Line — In Plain English
What it is
THC is the main psychoactive compound in cannabis. It works on the body's own endocannabinoid system (the CB1/CB2 receptors) — the same system involved in pain, appetite, mood, and memory.
What the research shows
Solid for chemo nausea and MS spasticity (approved synthetic versions exist). Modest for chronic pain and appetite. Weak/mixed for sleep and anxiety.
Who uses it & how
24 states + DC allow adult use; 40 + DC allow medical. The standing advice is start low, go slow — potency today is far higher than a generation ago.
What the law says (2026)
A 2026 federal order moved medical marijuana to Schedule III, but recreational cannabis is still Schedule I federally. State and federal law still don't fully agree.
The honest verdict
Real medicine in a few narrow lanes — where cheap generics often match it. The "cures everything" claims outrun the data. And the harms are real but dose-, potency-, and age-dependent.
- Approved THC medicines are narrow: chemo nausea and AIDS appetite (dronabinol/nabilone); MS spasticity (nabiximols, abroad).
- Its effect on chronic pain is real but small — about 1 in 5 get major relief vs. 1 in 6 on placebo.
- ~9% of people who ever use it become dependent — the lowest of the major substances, but ~1 in 6 if use starts in the teens.
- Daily high-potency use carries roughly 5× the odds of psychosis in susceptible people — potency, not just use, is the key variable.
- In every lane it treats, a proven generic (gabapentin, baclofen, ondansetron) runs $5–20/month vs. ~$300/month at a dispensary.
- The biggest research gap is a policy artifact — Schedule I throttled rigorous U.S. trials for decades.