Every figure below is scoped to the European Union: alcohol, tobacco, caffeine, cannabis, opioids, cocaine and amphetamines compared on EU usage, EU mortality and EU cause-of-death — plus a new modelled estimate of how much alcohol contributes to cancer deaths beyond the official count, through its interaction with tobacco.
Of the ~4.84 million deaths recorded in the EU in 2023, tobacco and alcohol together account for roughly 1 million — around 20%. All illegal drugs combined account for about 0.15%. Alcohol's true cancer toll is larger still once its interaction with tobacco is counted.
Legal, socially normalised substances dwarf illegal drugs in user numbers across the Union — by one to two orders of magnitude.
*Cocaine and ATS figures cover the 15–34 age bracket only — the full 15–64 EU total is not consistently published, so these bars understate the true all-adult user count and are not directly comparable in scale to the alcohol/tobacco/cannabis figures above.
Caffeine is almost certainly the single most-used psychoactive substance in the EU — plausibly a majority of its 450 million residents via coffee, tea and soft drinks. No EU agency publishes a rigorous region-wide user count, so it is excluded from the chart rather than estimated imprecisely.
4.84 million EU residents died in 2023. Here is the share attributable to each category of substance use.
Emerging risk: nitazenes — synthetic opioids up to 1,000× stronger than morphine — are implicated in a growing share of overdoses in Estonia and Latvia, per EUDA's early-warning system.
Substance-attributable deaths aren't a separate bucket from "real" diseases — tobacco and alcohol kill people through cardiovascular disease, cancer and respiratory disease. Ranked alongside the EU's other leading causes of death, they sit near the very top.
| Cause of death | EU deaths/year | Share of all EU deaths | Data year | Type |
|---|---|---|---|---|
| Circulatory diseases (heart disease, stroke, etc.) | ~1,588,000 | 32.8% | 2023 | ICD category |
| Cancer (all types) | ~1,157,000 | 23.9% | 2023 | ICD category |
| Tobacco (attributable — sits inside the categories above) | ~700,000 | 14.5% | 2023 | Risk factor |
| Respiratory diseases (non-cancer, e.g. COPD, pneumonia) | ~377,500 | 7.8% | 2023 | ICD category |
| Alcohol (attributable — sits inside the categories above) | ~291,000 | 6.0% | 2016–19 | Risk factor |
| External causes (accidents, suicide, homicide, falls) | ~242,000 | 5.0% | 2023 | ICD category |
| — of which road traffic | ~20,400 | 0.4% | 2023 | |
| — of which suicide | ~47,346 | 1.0% | 2021 | |
| Dementia, incl. Alzheimer's | ~177,000 | 3.7% | 2022 | ICD category |
| Diabetes mellitus | ~114,400 | 2.4% | 2016 | ICD category |
| All illicit drugs combined (attributable) | ~7,500 | 0.15% | 2023 | Risk factor |
Rows marked Risk factor are not a separate category from the ICD category rows — they cut across them. A tobacco-caused lung cancer death is counted once, inside "Cancer"; a tobacco-caused heart attack is counted inside "Circulatory diseases." So the table does not sum to 4.84 million, and it shouldn't. What it does show: if tobacco were its own disease category, it would rank third — above respiratory disease, above dementia, above every external cause of death combined, and about 93× larger than every illegal drug combined.
Lung cancer is ~80–90% smoking-attributable in Europe; oral cavity, pharynx and larynx cancers are the sites most affected by the alcohol×tobacco synergy detailed in the next section. Cancer-site figures are EU-27, 2020 (ECIS); the broad-category chart uses 2023 Eurostat totals — a small year mismatch, noted for transparency.
IARC (WHO's cancer research agency) classifies carcinogens by strength of evidence, not degree of risk. Only two substances on this list meet the highest bar — and they are the two the EU consumes most.
| Substance | IARC classification | Basis |
|---|---|---|
| Alcohol (ethanol in beverages) | Group 1 — Carcinogenic | Causes cancers of oral cavity, pharynx, larynx, oesophagus, liver, colorectum and female breast. Dose–response, no safe threshold. |
| Tobacco (smoked & smokeless) | Group 1 — Carcinogenic | Causes an estimated 19% of all cancers; lung, mouth, throat, bladder, pancreas, cervix and more. |
| Cannabis (smoked) | Not yet classified | Smoke contains similar combustion carcinogens (PAHs) to tobacco, but epidemiological cancer evidence is "suggestive, not conclusive." IARC's first formal review is scheduled for Nov 2026. |
| Opioids | No classification | No established carcinogenic mechanism; used medically at scale. |
| Cocaine / Amphetamines | No classification | No established carcinogenic mechanism; harm is overwhelmingly cardiovascular. |
| Caffeine / coffee | Group 3 — Not classifiable | Downgraded from a weaker 2B classification in 2016 after review of 1,000+ studies. |
Alcohol is not a single-organ risk. It is one of the few substances on this list that measurably damages nearly every organ system, with evidence strong enough that major health bodies have stopped describing any dose as "protective."
Fatty liver → alcoholic hepatitis → cirrhosis. Risk rises steeply above roughly 2 drinks/day sustained over years.
IARC Group 1 carcinogen — causally linked to cancers of the mouth, throat, oesophagus, liver, colorectum and breast. Risk rises with consumption and there is no threshold below which it is zero; even light/moderate drinking raises breast cancer risk in women. (Full modelled analysis below.)
Chronic heavy use raises blood pressure, causes cardiomyopathy, and increases stroke and atrial-fibrillation risk. The old "red wine is heart-healthy" claim has been substantially walked back: large cohort and Mendelian-randomization studies suggest the earlier observational link was confounded (sick-quitter effect, lifestyle factors). WHO and a growing number of national guidelines no longer endorse any protective dose.
Neurotoxic — impairs memory and cognition. Chronic heavy use is linked to reduced brain volume and higher dementia risk, plus peripheral neuropathy.
Dependence, worsened anxiety and depression, and disrupted sleep architecture — alcohol suppresses REM sleep despite its sedative onset effect.
Weight gain (7 kcal/g of "empty" calories), pancreatitis risk, and interference with blood sugar regulation.
Weakened immune function, plus materially increased risk of injury and accidents while intoxicated.
No known safe amount. Prenatal exposure carries a risk of fetal alcohol spectrum disorders (FASD).
Risk is roughly dose-dependent for most of the conditions above, but for cancer specifically it is non-linear and starts from zero. That distinction is the main shift in public-health messaging over the last several years: away from "moderate drinking is fine" and toward "less is better, none is best."
Official statistics credit alcohol with ~79,000 EU cancer deaths a year. That number is real — but it is built on a model that assigns each cancer death to one dominant risk factor. It structurally cannot capture what happens when alcohol and tobacco act together.
A peer-reviewed EU-27 study estimated 78,985 alcohol-attributable cancer deaths in 2016 (95% CI 75,498–91,102): 56,207 in men, 22,778 in women, and 1.9 million years of life lost. These deaths are drawn from the seven cancer sites IARC recognises as causally linked to alcohol: oral cavity, pharynx, larynx, oesophagus, colorectum, liver and female breast.
For three of those sites — oral cavity, pharynx and larynx — alcohol rarely acts alone. The INHANCE consortium's pooled analysis of head-and-neck cancer (Hashibe et al., Cancer Epidemiology, Biomarkers & Prevention, 2009) found the combined effect of alcohol and tobacco is greater than multiplicative (interaction ψ = 2.15, 95% CI 1.53–3.04) — the two substances amplify each other's damage rather than simply adding to it.
EU-27 records ~43,400 deaths a year from oral cavity, pharynx and larynx cancer combined (ECIS, 2020: 31,800 + 11,600). Applying the INHANCE breakdown to that total:
Roughly 15,000 EU deaths a year from head-and-neck cancers occur only because alcohol and tobacco are consumed together — deaths that a standalone "alcohol-attributable" statistic does not credit to alcohol, because in isolation (no smoking present) alcohol alone would explain only ~1,700 of these same deaths. Counting the synergy, alcohol is a necessary contributing cause in up to 39% of EU head-and-neck cancer deaths (~16,900/year) — roughly ten times higher than an "alcohol acting alone" model would suggest.
Two further mechanisms plausibly add to alcohol's cancer burden but lack a robust EU-wide quantitative estimate, so no number is given — flagging them is more honest than inventing one:
Between the official direct count (~79,000/year) and the modelled tobacco-synergy addition confined to head-and-neck cancers (~15,000/year), alcohol's plausible true cancer footprint in the EU is meaningfully larger than the headline statistic suggests — though the two figures are not simply additive: it is not established how much of the ~79,000 official figure already implicitly reflects some joint risk, since studies vary in whether they adjust for smoking. Treat ~79,000 as a firm floor, ~15,000 as a well-evidenced but separately-modelled additional exposure specific to head-and-neck cancers, and the true combined total as genuinely uncertain — plausibly in the ~85,000–95,000+ range, not a precise figure any agency currently publishes.
Dividing EU deaths by EU user counts for the same year reveals something the total death tolls hide: a person who becomes a high-risk opioid user faces almost exactly the same annual death rate as the EU's average smoker.
Tobacco: ~700,000 EU deaths ÷ ~117 million EU smokers. Opioids: ~5,175 EU opioid-involved deaths ÷ 860,000 EU high-risk users. Alcohol: ~291,000 EU deaths ÷ ~295 million EU drinkers. These are order-of-magnitude estimates dividing same-region, same-era figures — not epidemiologically adjusted for dose, age or years of use. Cocaine and ATS are omitted: EU-wide, all-adult user totals for these two are not consistently published, so no reliable denominator exists.
The most widely cited multi-criteria expert assessment of a European country scored 20 drugs 0–100 on harm to the user and harm to others (crime, family, economic cost). It is now 16 years old, but no comparably rigorous EU-wide update exists — and EU health bodies still cite it.
The full list of 20 assessed drugs also included (lower harm): benzodiazepines 15, ketamine 15, methadone 14, mephedrone 13, khat 9, ecstasy 9, LSD 7, mushrooms (psilocybin) 5. Caffeine was not assessed in this study.
| Substance | EU users | EU deaths/year | IARC classification | Legal status (EU) |
|---|---|---|---|---|
| Tobacco | ~117 million | ~700,000 | Group 1 | Legal, regulated |
| Alcohol | ~295 million | ~291,000 (~79,000 official cancer + ~15,000 modeled addition from tobacco synergy) | Group 1 | Legal, regulated |
| Caffeine | Most of ~450M population (est.) | Negligible | Group 3 | Legal, unregulated |
| Opioids (high-risk users) | 860,000 | ~5,175 | Unclassified | Illegal / prescription only |
| Cannabis | 24 million (15–64) | Not established | Pending (2026) | Varies by member state |
| Amphetamines / ATS | 1.6 million (15–34)* | Included in aggregate illegal-drug figures | Unclassified | Illegal |
| Cocaine | 2.7 million (15–34)* | ~1,050+ (rising trend)** | Unclassified | Illegal |
*Youth subgroup only (15–34), not comparable in scope to the whole-population figures above. **Subgroup of 20 reporting countries, not the full EU-27.
Figures are drawn from the latest release of each source (alcohol data reference 2016–2019; EUDA/ECIS data mostly 2020–2023) and are not perfectly time-aligned. Drug-induced death reporting varies by country and coroner practice, and likely undercounts polysubstance deaths. Cocaine/ATS-specific mortality cannot be cleanly separated from aggregate "illegal drugs" figures in EU statistics. The Nutt harm score is a 2010 UK expert estimate, not a death count. Cannabis carcinogenicity is a genuinely open question pending the 2026 IARC evaluation. The alcohol-tobacco synergy estimate (steps 2–3 above) is a transparent model built on published relative-risk data, not an official government statistic — treat it as indicative.
Ranked by potential lives saved per euro spent, not by political appeal.
Law enforcement and media attention in most EU states are heavily weighted toward illegal drugs, which cause less than 0.2% of substance-attributable deaths. Alcohol and tobacco cause the remaining 99.8% — and alcohol's true cancer burden, once the tobacco synergy is counted, is higher than the headline figure usually cited. Reallocating resources toward evidence-based harm is the single highest-impact structural change available.