Drug legalisation and crime: what changes, and what does not?

The evidence in one minute

Removing personal possession from criminal law is not the same as legalising production and sales. Neither makes drugs harmless. There is no defensible blanket percentage for the resulting crime reduction.

• Portugal did not legalise drug sales. Decriminalisation accompanied wider health measures; improvements cannot all be attributed to the law alone.
• Fewer offences do not necessarily mean fewer victims. Changing the law changes crime counts. Assault, robbery and organised crime require separate measurement.
• Legal alcohol and tobacco cause substantial harm. Comparisons depend on exposure and outcome. Cannabis findings cannot simply be transferred to cocaine, MDMA or opioids.
• The rules matter. Pricing, age limits, marketing, treatment and enforcement affect outcomes. Illegal markets can shrink, persist or shift; none is guaranteed.

Research date: 20 September 2026. AI-assisted research and writing; no independent clinical or criminological review. General information, not personal advice. Sources cover different observation periods, not necessarily the situation today.

1. “Legalisation” is not a single policy

Public debate often contrasts “banning drugs” with “allowing drugs”. Actual policy has several separate components: possession, production, distribution, retailing, advertising and treatment. A country can relax one component while retaining strict controls on another. These distinctions determine which benefits and risks are plausible. They also prevent a successful treatment trial being presented as evidence for unrestricted retail sales.[1]

Terminology varies between publications. Therefore, look beyond the label: who can obtain which product, from whom, and with what oversight? A public monopoly, a non-profit model and competitive commercial retailers can all be described as regulated legal supply, yet create different incentives. Age restrictions, product standards and marketing rules are substantive policy choices, not minor details.

Heroin-assisted treatment illustrates the importance of this distinction. A Cochrane review of 8 randomised studies involving 2,007 participants found benefits for treatment retention and reduced illicit drug use among people with longstanding dependence and previous treatment failures. Some studies also found less offending. Medication-related adverse events were more frequent; the mortality result was uncertain. The evidence supports a carefully supervised option for a selected clinical group, not the safety of an open heroin market.[10]

2. Alcohol and tobacco: comparisons need context

Alcohol and nicotine are psychoactive substances. Their legal status does not put them outside the health discussion. WHO attributes about 2.6 million deaths worldwide in 2019 to alcohol. Tobacco causes millions of deaths annually, including deaths from second-hand smoke. These are population burdens, not the probability that a particular occasion of drinking or smoking will be fatal.[2][3]

A fair comparison specifies the outcome: acute poisoning, dependence, cancer, cardiovascular disease, psychosis, injury or harm to other people. It also specifies exposure: amount, frequency, duration, route of administration, product composition and combinations with other substances. Comparing occasional use of one drug with prolonged heavy use of another does not establish a general safety ranking.

Individual risk and population burden are different. A widely used substance can cause substantial total harm even when the risk per exposure is lower on a particular measure. A less common substance can have a high risk for its users while accounting for fewer deaths nationally. Neither “more deaths overall” nor “fewer deaths overall” is a sufficient answer to whether a particular product or policy is safer.

Tobacco's major disease burden is strongly connected to its products and patterns of exposure, including smoke; nicotine dependence and smoke-related disease are not interchangeable concepts. Alcohol contributes to cancer, liver disease, injuries and violence. These examples support taking regulation seriously, rather than treating existing legal markets as an ideal template. WHO documents benefits from tobacco taxation and advertising restrictions, while illicit tobacco trade still exists.[2][3]

What about the famous drug-harm rankings?

Nutt and colleagues' British analysis, published in 2010, scored 20 substances against 16 criteria. Alcohol ranked highest overall in that exercise, whereas the ranking for harm to the individual differed. This was a structured expert assessment with explicit weights, not an experiment giving comparable exposures to participants.[4]

Such an exercise makes overlooked harms visible. But the overall score depends on judgements about how much weight to give dependence, death, family disruption, economic costs and other outcomes. It also reflects the social setting and existing markets. The result is not an objective, universal league table and does not show that a lower-ranked drug is harmless. Nor does it directly predict what would happen if that drug became more available.

3. What can the research actually establish?

Policy research rarely has the experimental control of a clinical trial. Countries choose reforms rather than being randomly assigned to them. Treatment provision, policing, the economy and retail rules often change at the same time. A before-and-after graph can reveal a trend, but not necessarily the cause. The hardest question is: what would have happened in the same place without the reform? We cannot directly observe that situation.

Comparison regions, interrupted time-series analyses and synthetic controls can improve causal inference. They still depend on assumptions: comparable underlying trends, no major unmeasured simultaneous change, and limited spillovers across borders. Researchers can reach different estimates by selecting different comparison areas, start dates or outcomes. A null result may mean little effect, but also insufficient data to detect one.

A systematic review by Scheim and colleagues, published in 2020 and searching studies through October 2018, included 114 articles. Of these, 109 concerned cannabis and 104 came from the United States. Many designs were relatively weak, and outcomes focused heavily on use rather than wider health and social effects. Its finding that many measures showed no change is not proof that every reform is harmless or ineffective.[1]

The evidence is therefore strongest for some narrower questions, such as changes in cannabis possession enforcement, purchasing patterns and selected treatment outcomes. It is much thinner for the net effect of legal retail cocaine or MDMA, or for nationwide changes in organised-crime violence. Cannabis is not a reliable substitute for studying drugs with different dependence patterns, acute toxicity, markets and consumer populations.

4. Portugal: decriminalisation, not legal drug shops

Portugal's reform took effect in 2001. Possession for personal use within its framework moved from criminal punishment to an administrative system, involving commissions that can assess the situation and direct people towards support. Unauthorised drug production and trafficking did not become lawful. “Portugal legalised all drugs” is therefore a misleading description.[5]

The reform belonged to a broader response involving treatment and harm reduction. Published assessments describe encouraging trends, including reductions in infectious disease and some criminal-justice burdens. They do not establish that the legal change alone caused all improvements. The epidemic was changing, services were changing, and different indicators cover different populations and periods.

A critical policy analysis published in 2021 also describes limitations: continuing stigma, uneven development of harm reduction and legal interpretations that complicated the simple story of “no criminal punishment”. That article analyses historical developments, not the complete legal position in 2026. It is useful precisely because it does not treat Portugal as an uncomplicated success story.[5]

The defensible lesson is that treating personal drug use primarily as a health and social issue can coexist with criminal enforcement against suppliers. It does not establish an optimal retail model or a guaranteed reduction in overdose deaths. Decriminalisation without accessible treatment, housing and other support should not be assumed to reproduce a broader policy package; equally, problems in service delivery do not by themselves prove that criminalising possession would improve health.

5. Canada: an important distinction between market share and harm

Canada legalised non-medical cannabis in 2018, with regulated production and sales. Its experience offers evidence about a legal cannabis market, not about legal retail sales of all drugs. Federal rules and provincial differences also mean that “the Canadian model” contains more than one retail arrangement.[7]

Official data show substantial substitution towards legal suppliers. However, several commonly quoted percentages measure different things. Spending, usual source, any legal purchase and exclusive legal purchasing are not interchangeable. Survey respondents may also misunderstand whether a seller is licensed.

The expert panel's 2024 report welcomed movement into the regulated market and fewer possession charges, while describing a persistent illicit market. It also highlighted high-potency products, accidental poisonings among children and increased cannabis-related health-care presentations. These concerns cannot be dismissed merely because legal purchasing increased.[7]

What about young people and overall health?

The panel described youth use as relatively stable after legalisation, but still high, and use among young adults as increasing. “Youth”, “young adults” and “all adults” are different groups. Stable prevalence also does not rule out changes in frequency, product strength or problems among a smaller group of frequent users. The cannabis survey itself warns that its recruitment methods may produce higher prevalence estimates than other Canadian surveys.[6][7]

Canada thus demonstrates both opportunities and constraints: meaningful legal-market substitution, continued illegal supply and unresolved health concerns. It does not provide a single “success percentage”. A programme can reduce criminal-justice harm while failing to reduce some medical harms, and both findings deserve to be reported.

6. The United States: mixed findings need not be contradictory

US states introduced different medical and recreational laws at different times. Permission to possess, the opening of retailers and later market expansion are separate events. A study of medical access is not automatically a study of commercial recreational supply. National averages can also conceal local effects.

Lu and colleagues' study, published in 2019, used comparison states and interrupted time-series methods to examine the early legalisation and retail-sales periods in Colorado and Washington. It found no statistically significant long-term effect on aggregate violent or property crime in those states. That argues against confident claims of either a general crime explosion or a large general crime reduction in that setting.[8]

A different study by Wu and Willits, published in 2022, used data from 2007–2017 and found an increase in recorded simple assault in Oregon counties relative to comparison states after recreational legalisation. It concerned a different outcome and setting, not the same test with the opposite answer. Neither study measures every form of victimisation or proves that cannabis caused a particular violent act.[9]

Health and safety findings are similarly outcome-specific. A 2024 systematic review of traffic-injury emergency visits and hospitalisations included 7 studies from Canada and the United States: 4 reported increases after legalisation and 3 found no significant change. The authors identified moderate overall risk of bias and could not pool results because studies differed. This is reason for careful monitoring, not a universal numerical prediction for Dutch roads.[18]

7. Why a falling crime count is not the whole answer

If conduct ceases to be an offence, recorded offences for that conduct can fall without any change in how often people do it. Canada's expert panel reports a 95% reduction in cannabis possession charges between 2017 and 2022. This is an important reduction in exposure to criminal proceedings, but it is not a 95% reduction in violence, drug use or organised crime.[7]

Calling this only a “statistical trick” is also unfair. Avoiding prosecution, a conviction and related disruption can be a real benefit. The point is to name the outcome accurately. The legal definition changes the count; the reduction in criminal-justice involvement can still matter to people.

The Dutch CBS Safety Monitor illustrates a different way of measuring crime: 20% of residents aged 15 and older reported experiencing traditional crime in 2023, compared with 17% in 2021. This is a survey measure of victimisation, not a police-recorded crime total, and not an evaluation of drug legalisation. “Traditional crime” here covers violence, property crime and vandalism; online crime is reported separately.[11]

Seizures and dismantled laboratories need similar caution. More discoveries may indicate a larger market, better detection or a changed investigative focus. Fewer discoveries may indicate either less activity or less detection. The Dutch National Drug Monitor explicitly describes fragmented seizure data and incomplete coverage. Counting discoveries alone cannot settle whether a drug policy succeeded.[15]

8. Violence has more than one connection to drugs

Intoxication-related violence concerns effects of a substance, interactions with other substances and the surrounding circumstances. Alcohol is an important example. Individual history, provocation and setting also matter. The Dutch National Drug Monitor stresses these contextual factors; toxicology among selected suspects is not a representative sample of every violent incident or proof of causation.[16]

Acquisitive offending concerns obtaining money to sustain costly use or dependence. Effective treatment and social support may reduce that pressure for some people. Findings from supervised heroin treatment are relevant here, but not evidence that all crime by people who use drugs is caused by the price of drugs. Housing insecurity, other dependencies and pre-existing offending can remain.[10]

Market violence concerns disputes over territory, debts, retaliation and control of illegal business. Regulated suppliers can in principle use contracts and courts rather than private coercion. Moving transactions out of illegal markets could therefore reduce one source of violence. But violence can persist in residual illegal markets, around other substances or in unrelated disputes. Legal status does not remove every motive for intimidation or robbery.

A 2011 systematic review by Werb and colleagues found that most included studies associated intensified drug-law enforcement with more market violence. Disruption may create unstable competition, but the predominantly observational evidence does not show that every police intervention causes violence, or that every legalisation model prevents it. Targeting violence and corruption differs from indiscriminate market disruption. The mechanism matters more than the slogan “more policing” or “less policing”.[12]

9. Displacement: neither automatically complete nor absent

“Criminals will simply do something else” identifies a real possibility, not a measured law of nature. People, networks and assets can move into other illegal activities. But another market must offer customers, opportunities and workable margins; competition and barriers to entry do not disappear. Losing a profitable market can therefore reduce criminal income even if some participants adapt.

Separate market substitution from crime displacement. A consumer switching to a licensed cannabis shop is the former. A former supplier moving into extortion or another drug market is the latter. A legal share can rise while the illegal market's absolute size stays stable if total demand expands. Revenue can also fall through lower prices without a proportionate reduction in physical volume.

Residual illegal supply is especially plausible where legal prices are unattractive, authorised shops are inaccessible or consumers seek excluded products. Taxes and compliance costs can affect the price gap. Age limits deliberately leave minors outside legal retail access. Yet lowering taxes or expanding availability to compete with illicit sellers can conflict with health protection. Canada's expert panel describes both displacement of illegal trade and continuing illicit operators; neither a completely eliminated market nor no displacement fits that account.[7]

To measure net displacement, researchers need to follow people, markets, revenues, violence and neighbouring areas over time. Counting a new offence elsewhere is insufficient: it may have increased anyway. Assuming complete replacement overstates offenders' flexibility; assuming none ignores it. For Dutch policy, the honest answer is that the proportion is uncertain and likely varies by substance, network, geography and timeframe.

10. What is realistically at stake in the Netherlands?

The Netherlands did not historically legalise the entire cannabis supply chain. Coffeeshop retail sales are tolerated under conditions, while supply outside authorised arrangements remains illegal. That is the familiar backdoor problem: tolerated retailing coupled to unlawful wholesale production and supply. Describing the whole system as legal obscures the very problem that supply-chain reform seeks to address.[13]

The controlled cannabis supply-chain experiment is a specific exception, not general legalisation of drugs. The government's timeline records the experimental phase beginning on 7 April 2025, with participating coffeeshops supplied by designated growers. Enforcement against unregulated hashish was temporarily deferred until 1 September 2025. These are implementation milestones, not findings that health improved or organised crime declined. The experiment concerns participating municipalities and cannabis, not a nationwide cocaine or MDMA market.[14]

Regulating supply could improve traceability, product oversight and accountability while reducing participating shops' reliance on illegal suppliers. Whether this translates into less total offending requires evaluation outside the shops as well. Changes in neighbouring areas, consumer sourcing, product characteristics and health outcomes are relevant. A process working as intended is not yet proof of a favourable overall outcome.

The wider Dutch drug economy includes international trafficking and synthetic-drug production for foreign customers. The National Drug Monitor documents cross-border supply, synthetic-drug laboratories and waste dumping. Reforming domestic cannabis sales does not remove foreign demand for illicit drugs. Even wider domestic regulation would not automatically legalise exports or eliminate foreign criminal markets. No defensible percentage of all Dutch organised crime can be promised from cannabis reform alone.[15]

Regulating composition does not remove a drug's own toxicity. For example, the National Drug Monitor describes serious MDMA complications, including overheating and disturbances of salt balance, even without assuming an adulterated product. The case for testing or production standards must therefore be separated from a claim of harmlessness. This report provides no basis for recommending non-medical consumption.[17]

11. The strongest reasonable arguments on both sides

These arguments do not force a choice between unrestricted commerce and maximum punishment. A serious assessment compares concrete options, including maintaining current rules, and specifies goals before measuring results. Monitor health, victimisation, criminal-justice burdens, illegal-market activity and distributional effects separately. The evidence supports some narrower conclusions; it does not settle how society should weigh liberty, protection, fairness and risk, or justify a promised numerical reduction in violence.

12. Methods, source access and limitations

This is a sourced explanatory report, not a new systematic review or a complete legal inventory. Research was checked on 20 September 2026. Sources were selected for direct relevance: official statistics and policy descriptions, systematic reviews and original studies. Publication dates and observation periods are distinguished where figures are used. Older sources provide historical evidence, not measurements of the situation on the research date.

The Canadian expert review combines evidence with consultation and recommendations; it is not a controlled experiment. The Portuguese article is a critical legal and policy analysis, not a causal impact trial. The National Drug Monitor compiles official and research data and flags gaps in registration. The harm-ranking, Oregon assault, enforcement-violence and traffic-review findings below were checked against indexed abstracts; paywalled full texts were not independently assessed. Their limitations should not be hidden by the presence of a DOI.

Numbers are included only where their definition and source could be verified. No estimate is offered for the percentage of Dutch violence that legalisation would prevent. Uncertainty about that figure is not proof of no benefit, nor permission to assume a large benefit.

1. Scheim AI et al. Impact evaluations of drug decriminalisation and legal regulation on drug use, health and social harms: a systematic review. BMJ Open, 2020;10:e035148. Search through October 2018; full text accessed.
2. World Health Organization. Alcohol. Official fact sheet; the mortality estimate cited here concerns 2019, not 2026.
3. World Health Organization. Tobacco. Official fact sheet on health harms, regulation and illicit trade; accessed as a live page.
4. Nutt DJ, King LA, Phillips LD. Drug harms in the UK: a multicriteria decision analysis. The Lancet, 2010. Expert scoring, not comparative clinical risk measurement. Verified abstract and metadata via Europe PMC.
5. Rêgo X et al. 20 years of Portuguese drug policy — developments, challenges and the quest for human rights. Substance Abuse Treatment, Prevention, and Policy, 2021. Full-text historical policy analysis; not a current legal guide or causal trial.
6. Health Canada. Canadian Cannabis Survey 2024: Summary. Official survey; usual source, spending and prevalence have different denominators. Self-report and recruitment limitations apply.
7. Health Canada, independent expert panel. Legislative Review of the Cannabis Act: Final Report of the Expert Panel. 2024. Official review including the Statistics Canada spending estimate for the third quarter of 2023 and possession charges for 2017–2022.
8. Lu R et al. The Cannabis Effect on Crime: Time-Series Analysis of Crime in Colorado and Washington State. Justice Quarterly, published online 2019. Original quasi-experimental analysis; full text accessed.
9. Wu G, Willits DW. The Impact of Recreational Marijuana Legalization on Simple Assault in Oregon. Journal of Interpersonal Violence, published online 2022; data from 2007–2017. Verified abstract and metadata via Europe PMC; publisher full text not assessed.
10. Ferri M, Davoli M, Perucci CA. Heroin maintenance for chronic heroin-dependent individuals. Cochrane Database of Systematic Reviews, 2011;CD003410. Review summary and detailed abstract accessed; selected treatment-refractory populations, not retail legalisation.
11. Statistics Netherlands (CBS). Safety Monitor 2023, chapter on traditional crime. Published 2024. Population victimisation survey; not a drug-policy evaluation.
12. Werb D et al. Effect of drug law enforcement on drug market violence: a systematic review. International Journal of Drug Policy, 2011. Predominantly observational evidence. Verified abstract and metadata via Europe PMC.
13. Government of the Netherlands. Gedoogbeleid softdrugs en coffeeshops. Official explanation of tolerated retail sale; read alongside the experiment-specific rules.
14. Government of the Netherlands. Tijdlijn experiment gesloten coffeeshopketen. Official implementation timeline, including the experimental phase and hashish transition in 2025; no final impact evaluation is claimed.
15. National Drug Monitor, Trimbos Institute and WODC. Handhaving en bestrijding: inbeslagnames en ontmantelingen. Institutional monitoring of Dutch trafficking and production; includes registration limitations and data covering different years.
16. National Drug Monitor. Middelengebruik en geweld. Research synthesis and selected forensic registrations; explicitly notes the limited availability of recent Dutch research.
17. National Drug Monitor. Ecstasy: ziekte en sterfte. Institutional synthesis of acute and longer-term health risks; not evidence that regulated retail access is safe.
18. Dion PM et al. Road hazard: a systematic review of traffic injuries following recreational cannabis legalization. Canadian Journal of Emergency Medicine, 2024; search through October 2023. Verified abstract and metadata via Europe PMC; no pooled causal estimate.
19. Dutch police. TERUGLEZEN: Dode bij schietincident Overasselt, twee agenten gewond. 1 September 2026. Primary source for the context; not evidence about motive, legal guilt or the effects of drug policy.