The most comprehensive cannabis review ever found strong evidence for pain relief and psychosis risk — and insufficient evidence for most other claimed benefits.
Policymakers, clinicians, patients seeking evidence-based guidance on cannabis use, anyone wanting to know what the science actually shows.
468 pages, ~100 conclusionsThe most comprehensive cannabis evidence review ever assembled, finding strong evidence for 3 therapeutic uses and substantial evidence for several harms
The Backstory
In January 2017, with recreational cannabis legal in eight states and medical cannabis in twenty-eight, the United States government did something it should have done decades earlier: it asked a panel of scientists to review the evidence and tell it what cannabis actually does.
The National Academies of Sciences, Engineering, and Medicine (NASEM) assembled a 16-member expert committee chaired by Marie McCormick of Harvard. They reviewed more than 10,000 scientific abstracts. They spent months sorting evidence into categories — from "conclusive" to "insufficient." The result was a 468-page report that became the reference document for every subsequent policy decision about cannabis in America.
What it found was as notable for its gaps as for its conclusions.
What This Report Is (and Isn't)
This is not a single study. It's a consensus report — the closest thing the American scientific establishment has to an official verdict on a question. NASEM committees are convened by the National Academies at the request of government agencies (in this case, prompted by changing state laws and the need for federal guidance). The committee members are chosen for expertise, screened for conflicts of interest, and their conclusions require majority agreement.
The report covers every major health question about cannabis: therapeutic benefits, cancer risk, cardiovascular effects, respiratory effects, immune function, prenatal exposure, psychosocial effects, mental health, addiction, and more. For each, the committee assigned one of four evidence categories based on the strength and consistency of available research.
The Evidence Map
The Three Big Surprises
1. Chronic pain was the strongest therapeutic finding. Not epilepsy, not PTSD, not anxiety. Chronic pain — the condition that affects the most people and intersects most directly with the opioid crisis. This finding became a key argument for medical cannabis advocates and influenced the opioid-cannabis substitution discussion.
2. Epilepsy was rated "insufficient." This seems jarring now, given Epidiolex's FDA approval just 18 months later. But the NASEM committee was reviewing evidence available through 2016, before the definitive Dravet syndrome trials were published. It's a reminder that evidence categories change as research advances.
3. Most claimed benefits had insufficient evidence. For all the conditions people use cannabis to treat — PTSD, depression, IBS, Parkinson's, dementia, glaucoma, opioid dependence — the evidence was rated as limited or insufficient. Not because cannabis doesn't work for these conditions. Because nobody had run the trials to find out.
The Research Barrier Problem
The report's most impactful recommendation may have been its simplest: remove the barriers to cannabis research.
Schedule I
classification has been the single largest obstacle to cannabis research in the United States for fifty years. Researchers need DEA approval, a special license, and can only use NIDA-supplied cannabis — which until recently came from a single farm at the University of Mississippi.
Researchers studying heroin, methamphetamine, or fentanyl face fewer administrative barriers than researchers studying cannabis. The scheduling paradox has produced a bizarre situation: a substance used by 50+ million Americans is among the hardest drugs to study scientifically.
NASEM (2017), Chapter 15
The committee was explicit: the evidence base was thin not because cannabis was understudied relative to its scientific interest, but because regulatory barriers made rigorous research nearly impossible. The Schedule I classification — which assumes no medical value and high abuse potential — was itself preventing the research that could determine whether cannabis has medical value.
This recommendation has been only partially implemented. The DEA expanded research cannabis suppliers in 2021. Rescheduling to Schedule III was proposed in 2024. But as of 2026, the regulatory environment remains substantially more restrictive for cannabis research than for most other substances.
What's Changed Since
The NASEM report was a snapshot of evidence through 2016. Since then:
- Epilepsy has moved from "insufficient" to proven, with Epidiolex FDA approval
- Chronic pain evidence has grown but remains complex — whole-plant cannabis trials still lag behind
- Psychosis risk has been further strengthened by Di Forti's 2019 EU-GEI study
- The opioid relationship has become more nuanced, with population-level data and individual-level studies both supporting cannabis-as-substitute effects
- Sleep remains complicated — the evidence is still mixed
The core finding hasn't changed: for most conditions, the evidence base is thinner than the public assumes. Cannabis research has expanded since 2017, but the gap between what millions of users believe and what controlled trials have demonstrated remains enormous.
Key Takeaways
The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research
National Academies of Sciences, Engineering, and Medicine (2017) · National Academies Press
Does this report prove cannabis works for chronic pain?
The committee found "conclusive or substantial evidence" that cannabis is effective for chronic pain — their strongest therapeutic rating. This doesn't mean cannabis works for all pain types or all people, but the overall evidence was strong enough for scientific consensus. The evidence has continued to grow since 2017.
Why was so much rated 'insufficient'?
Primarily because of research barriers. Cannabis's Schedule I classification made rigorous clinical trials extremely difficult to conduct. The committee wasn't saying cannabis doesn't work for these conditions — it was saying nobody had done the research to find out. The report explicitly called for removing these barriers.
Is this report still relevant?
The evidence categories for some conditions have shifted (epilepsy most dramatically), but the report's core framework remains the reference standard. No comparable comprehensive review has been published since. Many of its "insufficient" ratings remain accurate because the research still hasn't been done.
Can I use this report to justify medical cannabis use?
For chronic pain, chemotherapy nausea, and MS spasticity — yes, the evidence is strong. For most other conditions, the evidence is weaker than many people assume. This report is useful precisely because it's honest about what we know and don't know.
What the researchers found
The most comprehensive cannabis evidence review ever assembled found conclusive evidence for three therapeutic uses (chronic pain, chemotherapy nausea, MS spasticity), substantial evidence for several harms (motor vehicle crashes, lower birth weight, psychosis risk), and insufficient evidence for most other claimed benefits. Called for removal of Schedule I research barriers.
Why it matters
This report became the reference document for cannabis policy worldwide. It revealed that for most conditions people use cannabis to treat, the evidence base is remarkably thin — not because cannabis doesn't work, but because research barriers prevented the trials from being conducted.
The numbers in context
- 16-member expert committee
- 10,000+ scientific abstracts reviewed
- ~100 research conclusions
- 468 pages
- 3 therapeutic uses with conclusive/substantial evidence
- Numerous conditions rated as insufficient evidence
How the study worked
Consensus committee convened by NASEM. Systematic review of all available evidence through 2016. Conclusions required majority committee agreement. Evidence categorized as conclusive/substantial, moderate, limited, or insufficient.
What this study cannot tell us
Evidence through 2016 only — some categories (especially epilepsy) changed rapidly after publication. Committee structure may not capture all perspectives. Consensus process can be conservative.
How to read the evidence
Authoritative: NASEM consensus report represents the highest level of evidence synthesis. Committee members screened for conflicts of interest. Conclusions require majority agreement.
The bigger picture
This report exposed a paradox: a substance used by 50+ million Americans had remarkably thin clinical evidence for most uses, primarily because Schedule I classification made research nearly impossible. The biggest finding may not have been any single conclusion but the systematic documentation of how little we actually know.
Read the original research
The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research
National Academies Press, Washington, DC
Published by the National Academies Press, the publishing arm of the US National Academies.
Citation
National Academies of Sciences, Engineering, and Medicine. (2017). The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research. National Academies Press, Washington, DC. https://doi.org/10.17226/24625