← All posts
alcoholsmokingmortality data

Alcohol vs Smoking: Which Kills More People Every Year?

Smoking kills around 8.7 million people per year globally. Alcohol kills around 3 million. But the comparison is more complicated than the headline numbers suggest — here is what the data actually shows.

|13 min read
Share

Picture a pub somewhere in Britain on a Saturday night. A man is at the bar with a pint of lager and a cigarette tucked behind his ear, waiting for his chance to go outside. He is, in the estimation of most public health frameworks, doing two of the most dangerous legal things a person can voluntarily do to themselves. But if you asked him which one was worse, he probably could not tell you with any precision.

Neither can most people. The two substances kill differently — through different mechanisms, on different timescales, affecting different organs — and they are rarely placed in direct comparison. When they are, the results are illuminating.

The live death counter ticks up by one every second. Roughly one of those deaths every four seconds is attributable to tobacco. One every ten seconds is attributable to alcohol. Both are slow, quiet killers, and both are legal almost everywhere.

The headline comparison

SubstanceAnnual deaths globally (WHO, 2024 estimates)Share of all deaths
Tobacco (all forms)~8.7 million15.4%
Alcohol~3.0 million5.3%
Both combined~11.7 million~20%

Tobacco kills roughly three times as many people as alcohol every year. That gap is real and large. But it requires context on both ends.

The tobacco figure includes all forms: cigarettes, cigars, pipes, smokeless tobacco, and — increasingly — second-hand smoke deaths. If you count only direct smoking deaths, the number is closer to 7 million. The 8.7 million figure includes those exposed involuntarily to other people's smoke.

The alcohol figure is also likely an undercount. Many alcohol-attributable deaths — particularly from cardiovascular disease, liver cancer, and traffic accidents — are not coded as alcohol-related in official records. Some researchers estimate the true figure is closer to 3.5-4 million when these attribution gaps are closed.

Our dedicated articles cover both in detail: how many people die from smoking every day and how many people die from alcohol every day.

The daily breakdown

Putting the annual numbers on a human timescale makes them more concrete.

SubstanceDeaths per yearDeaths per dayDeaths per hourDeaths per minute
Tobacco8,700,00023,83699316.6
Alcohol3,000,0008,2193425.7

Tobacco kills the equivalent of a fully loaded Boeing 747 every single hour, around the clock. Alcohol kills the equivalent of a small regional aircraft every hour. These are not abstract numbers — they represent people who were alive at the start of the year and will not be at the end of it.

How each substance kills

The most important thing to understand about this comparison is that the two substances kill through almost entirely different mechanisms, at different time delays, with different dose-response curves.

Tobacco

Tobacco is primarily a long-latency carcinogen and vascular toxin. Most of its killing happens through cancer and cardiovascular disease — processes that take years or decades to develop. A 60-year-old dying of lung cancer today may have developed that cancer from smoking he did in his 30s and 40s.

Cause of deathAnnual tobacco-attributable deaths globally
Lung cancer~1,800,000
COPD (chronic obstructive pulmonary disease)~1,200,000
Ischaemic heart disease (heart attacks)~1,700,000
Stroke~750,000
Other cancers (mouth, throat, oesophagus, stomach, bladder, kidney)~900,000
Lower respiratory infections~350,000
Other cardiovascular~700,000
Other causes~300,000

Lung cancer is the most visible — smoking causes approximately 85% of all lung cancer cases globally. But cardiovascular disease is quantitatively larger: smoking causes an estimated 1.7 million heart attack deaths and 750,000 stroke deaths per year. The effect is through multiple mechanisms: accelerating atherosclerosis, raising blood pressure, reducing HDL cholesterol, thickening blood, and causing inflammation throughout the vascular system.

Our article on how many people die of heart attacks per day covers the cardiovascular toll in detail. A meaningful fraction of every figure in that article traces back to tobacco.

Alcohol

Alcohol kills through a wider and in some ways more varied set of mechanisms. It is a toxin, a carcinogen, a central nervous system depressant, and the proximate cause of millions of accidents and acts of violence. Its death toll is distributed differently across age groups than tobacco's.

Cause of deathAnnual alcohol-attributable deaths globally
Liver disease (cirrhosis, alcoholic hepatitis)~500,000
Cardiovascular disease~500,000
Road traffic accidents (drunk driving)~400,000
Cancer (breast, colorectal, liver, mouth, oesophagus)~400,000
Intentional self-harm (suicide)~280,000
Alcohol poisoning / acute toxicity~180,000
Drowning and other unintentional injury~170,000
Violence (homicide)~220,000
Other~350,000

The alcohol distribution is more front-loaded in terms of age. Road accidents, violence, and acute poisoning kill younger people. Liver disease and cancer kill people in middle age. Tobacco's cardiovascular and cancer deaths are more concentrated in older age groups. If you compare years of life lost rather than raw death counts, alcohol's relative disadvantage narrows compared to the raw numbers.

The violence component is worth noting. A significant fraction of all homicides worldwide are alcohol-attributable — the WHO estimates 26% of all violence-related deaths involve alcohol. This is a category of mortality that tobacco essentially never causes directly.

The dose-response comparison

Both substances have dose-response relationships — more use means more risk — but the shapes of those curves are different.

Tobacco dose-response

Smoking statusRelative risk of death vs never-smokerLife expectancy reduction
Never smoker1.0 (baseline)
Light smoker (under 10/day)2.0-2.5×4-6 years
Moderate (10-19/day)2.5-3.0×8-10 years
Heavy (20+/day)3.0-4.0×10-14 years
Very heavy (40+/day)4.0-5.0×13-17 years

There is no safe level of smoking. Even occasional smokers — under 1 cigarette per day on average — show elevated mortality risk compared to non-smokers. The dose-response is steep and does not level off at any point studied.

The life expectancy reduction of 10-14 years for a pack-a-day smoker is one of the most robust findings in epidemiology. Smoking is, per unit of daily consumption, among the most deadly legal activities a person can engage in.

Alcohol dose-response

The alcohol dose-response is more complicated because of a phenomenon that has generated enormous debate: the potential protective effect of light drinking on cardiovascular disease.

Daily alcohol consumptionAll-cause mortality vs abstainers
0 units (abstainer)Baseline
Under 1 unit/day (occasional)Possibly slightly lower (disputed)
1-2 units/day (light)Roughly similar to abstainers
2-4 units/day (moderate)Modestly elevated
4-6 units/day (heavy)30-50% higher
Over 6 units/day (very heavy)2-3× higher
Binge pattern (regardless of weekly total)Elevated even at low weekly volumes

The "J-curve" or "U-curve" for alcohol — the idea that light drinkers have lower mortality than abstainers — was a major finding in 1990s and 2000s epidemiology. However, more recent research using Mendelian randomisation (a way to remove confounding) has raised serious doubts about whether this effect is real or an artefact of abstainer populations including former heavy drinkers with pre-existing health conditions.

The current scientific consensus has shifted: there is probably no cardiovascular benefit to light drinking that would not be achieved equally well by other lifestyle factors. The WHO now states that "no amount of alcohol is completely safe." That does not mean one drink a week is catastrophic — the absolute risk at very low doses is very small — but the protective effect that was once claimed is no longer considered established.

Where the deaths are concentrated geographically

Both substances kill unevenly around the world, for different reasons.

RegionTobacco deaths per 100,000 populationAlcohol deaths per 100,000 population
Eastern Europe200-25050-90
East Asia180-22020-30
Western Europe120-16020-35
North America130-15020-35
South Asia150-18010-20
Sub-Saharan Africa40-6025-45
Latin America60-10035-55
Middle East40-70under 5 (religious prohibition)

Eastern Europe stands out for both. Countries like Russia, Ukraine, and Belarus have among the highest tobacco and alcohol mortality rates in the world. Russia's extreme alcohol mortality rate — which has historically contributed to significantly lower male life expectancy — is well-documented. Our life expectancy by country article shows how dramatically Russian male life expectancy lags behind comparable income-level nations.

The Middle East pattern is informative: where alcohol consumption is low due to religious and cultural norms, alcohol mortality is essentially negligible. But tobacco rates remain elevated, suggesting these two substances operate independently at the population level — there is no simple substitution effect between them.

Quitting: the mortality reversal

Both substances show clear mortality benefits from cessation — but the timescales differ.

Tobacco cessation

Time after quittingBiological change
20 minutesHeart rate and blood pressure drop
12 hoursCarbon monoxide levels normalise
1 yearExcess cardiovascular risk halved vs smokers
5 yearsStroke risk drops to near-non-smoker levels
10 yearsLung cancer risk halved vs continuing smokers
15 yearsCardiovascular risk similar to never-smoker

The 15-year figure is remarkable: a person who smoked heavily for 20 years and then quits can have near-equivalent cardiovascular risk to someone who never smoked. The damage is not fully reversible — lung cancer risk never returns entirely to baseline — but the trajectory is strongly positive. Every year of not smoking is a year of recovery.

Alcohol cessation

For heavy drinkers, cessation also reverses significant mortality risk:

Time after quittingLiver / health change
Days-weeksWithdrawal (medical risk for dependent drinkers)
1 monthLiver inflammation starts to reduce
6 monthsFatty liver often reverses in early-stage cases
1-2 yearsBlood pressure, sleep, and cognitive function improve
5 yearsCirrhosis risk in early-stage fibrosis significantly reduced
10+ yearsCancer risk starts to approach non-drinker levels

The key qualifier is that alcohol-related liver damage has stages. Fatty liver and early fibrosis are largely reversible. Established cirrhosis is not — it can be managed but the structural damage to the liver is permanent.

The combined effect

An important but underappreciated fact is that smokers who also drink heavily have dramatically worse outcomes than either behaviour alone would predict.

BehaviourRelative risk of head and neck cancer vs non-user
Non-smoker, non-drinker1.0 (baseline)
Smoker only5-7×
Heavy drinker only3-5×
Smoker and heavy drinker25-40×

The interaction for head and neck cancers (oral, pharyngeal, laryngeal) is multiplicative, not additive. Smoking and alcohol together are vastly more carcinogenic in the upper digestive tract than either alone. A person who both smokes and drinks heavily faces a cancer risk roughly 30 times higher than neither. The two substances appear to work synergistically — alcohol makes mucosal tissue more permeable to tobacco carcinogens, while tobacco suppresses the immune surveillance that might otherwise catch early cancers.

Our article on cancer deaths shows the full scale of cancer mortality. The tobacco-alcohol interaction explains a meaningful portion of the cancers of the mouth, throat, and oesophagus that contribute to that total.

Which is worse?

Smoking kills more people annually — by a factor of roughly three. That is not a close comparison at the global level. But "worse" depends on the frame.

Per user, heavy alcohol use may match tobacco in harm. The alcohol death rate is lower largely because fewer people drink at dangerous levels than smoke at dangerous levels — but a person who consumes 8 units daily is taking on substantial mortality risk that is comparable in magnitude to a heavy smoker.

In terms of years of life lost, alcohol's younger victims — accident deaths, violence, overdose — mean the per-death loss of life years is higher for alcohol than tobacco in many analyses. A 25-year-old killed in a drunk-driving accident loses 50 potential years. A 70-year-old dying of lung cancer loses far fewer.

Neither comparison is an argument for using the other. Both are significantly lethal. Both are preventable. And both are contributing to the counter you can watch right now on our live death counter.

While you have been reading this — roughly 13 minutes — approximately 780 people have died worldwide. Around 200 of them died from tobacco-related causes. Around 70 from alcohol-related causes. You can see the broader toll on our causes of death pages.


Data sources: World Health Organization Global Status Report on Alcohol and Health 2023, WHO Report on the Global Tobacco Epidemic 2023, Global Burden of Disease Study 2021, International Agency for Research on Cancer, The Lancet Global Health alcohol meta-analyses. This article is for educational purposes only and does not constitute medical advice.

Share

You might also like