Cannabis Legalisation Outcomes

Two dozen US states, Canada, Uruguay, Germany (partially) and others have legalised recreational cannabis, generating the largest drug-policy natural experiment ever run. The interim scoreboard confounds both camps' predictions. Not materialised: sustained increases in adolescent use (surveys show flat-to-declining teen use post-legalisation in most jurisdictions), crime waves, or carnage-scale road tolls — though driving impairment detection remains genuinely unsolved. Materialised: substantial increases in adult use — especially daily and near-daily use, which now exceeds daily alcohol use in US surveys; cannabis use disorder scaling with that heavy-use growth; emergency presentations for hyperemesis, paediatric edible ingestions and acute psychiatric events; and an illicit market that shrank less than promised while a commercial industry acquired the lobbying reflexes of its alcohol and tobacco elders. Legalisation neither freed nor doomed anyone on schedule; it industrialised a drug, with everything that implies.

The single most important fact in modern cannabis debates: the product changed. Flower THC content rose from low single digits in the 1990s to 15-25% in dispensary products, with concentrates at 60-90% — the 10-to-20-times-stronger framing US officials now use — meaning the reassuring older epidemiology measured a substantially different exposure. Potency matters because the serious harms are dose-dependent: high-THC products associate with higher rates of cannabis use disorder (roughly three in ten regular users develop it by standard criteria), more severe hyperemesis, and — most consequentially — stronger psychosis associations. Markets, meanwhile, reward potency (price per THC unit), and few legal regimes cap it: a structural collision between commercial logic and the dose-response evidence that regulation has barely begun to address.

The association between cannabis and psychosis is robust: heavy, high-potency, adolescent-onset use associates with several-fold increased psychotic-disorder risk in case-control and cohort studies, with dose-response gradients, and population-attributable estimates in high-potency markets reaching double-digit percentages of first-episode cases. What remains contested is the causal share: genetic confounding runs both directions (psychosis liability predicts cannabis use), self-medication of prodromal symptoms inflates associations, and — the strongest sceptical exhibit — population psychosis rates have not tracked the massive rise in use and potency the causal reading might predict, with several legalisation studies (state-level analyses, Canadian emergency-department series) finding no clear post-legalisation psychosis jump. The defensible synthesis: causality is probable for a vulnerable minority — adolescent-onset heavy users of high-potency products, especially with family history — while cannabis is neither necessary nor sufficient for psychosis and the marginal effect of legalisation per se on psychosis remains unproven. Both a real risk and an over-claimed one.

The US federal position shifted in stages: a 2025 executive order directed expedited rescheduling, and in April 2026 the Department of Justice moved FDA-approved cannabis products and medical cannabis under qualifying state licences to Schedule III — leaving recreational and non-qualifying cannabis in Schedule I — with a further DEA administrative process (hearings from June 2026) contesting broader rescheduling into 2027. The practical stakes are research access, banking and the 280E tax burden more than street-level change. Against this policy momentum stands an uncomfortable evidence audit: 2026 reviews in major journals reaffirm that rigorous support exists for only a short list of indications (certain childhood epilepsies via approved cannabidiol, chemotherapy nausea, MS spasticity, modest chronic-pain effects), while evidence remains weak-to-absent for the anxiety, depression, PTSD and insomnia indications driving most dispensary traffic — and essentially no psychiatric indication has trial support. Rescheduling fixes the research bottleneck that partly caused this gap; it does not retroactively validate the menu.

A decade of legal markets produced a curriculum of correctable mistakes: edibles engineered and packaged in ways children eat (paediatric ingestion calls rose sharply until packaging rules tightened); potency unlimited in most regimes while taxes were levied by weight rather than THC; marketing and retail density concentrating in lower-income neighbourhoods; delta-8 and synthetic-adjacent cannabinoids exploding through a hemp-law loophole into gas stations, unregulated and untested — arguably the worst products on the market wearing the least oversight; and expungement and social-equity provisions lagging the commerce they were promised alongside. The emerging better-practice list is unglamorous: THC-based taxation, potency caps or warnings, plain packaging and dosage limits for edibles, closing the hemp loophole, impaired-driving research investment, and treating cannabis-use-disorder services as part of the legalisation deal rather than an embarrassment to it.

For adult users in legal markets: the dose-response evidence converts into usable rules — favour lower-THC products (the difference between 10% and 80% is not a nuance), avoid daily use (dependence and hyperemesis live there), never drive impaired, and store edibles like the medicines they chemically are. Absolute avoidance is the evidence-based position for: adolescents (the developing-brain and psychosis window), pregnancy, personal or family psychosis history, and anyone noticing loss of control — three in ten regular users is not someone else. For patients considering medical use: match the claim to the evidence tier — real for a short list, unproven for most of the menu — and involve a clinician, especially regarding drug interactions. For voters and policymakers: the honest lesson of the experiment so far is that the interesting question was never legal-versus-illegal but what kind of market — and the public-health version (potency-aware, marketing-limited, treatment-funded) has to be legislated deliberately, because the commercial version arrives by default.