Smoking and alcohol
One is the clearest finding in public health. The other has spent a decade moving in the opposite direction from what people expected.
settledLarge, consistent evidence base. Unlikely to reverse.Revised 04 Sept 2026
The WHO's 2023 position: no level of consumption is safe for health.Photograph: CHUTTERSNAP / Unsplash
Smoking
There is no interesting scientific argument left here. Tobacco is a cause of cancer at multiple sites, of cardiovascular and respiratory disease, and it appears on the Lancet Commission’s list of modifiable dementia risk factors.
What is worth stating, because it is less well known, is that stopping produces measurable benefit at every age, including in people who have smoked for decades. Risk does not return to that of a never-smoker for many years, and for some outcomes never fully, but the curve bends within months.
For someone reading a reference on ageing well: this is the largest single return available. Nothing else on this site comes close.
Alcohol
The direction of travel has been consistent and uncomfortable. In January 2023, the WHO’s European office stated plainly that no level of alcohol consumption is safe for health — the risk of several cancers begins from the first drop, and there is no threshold below which the carcinogenic effect ceases.
Excessive alcohol consumption also sits on the Lancet Commission’s dementia list, alongside its established contributions to liver disease, hypertension and injury.
What happened to the J-curve
For years, observational studies appeared to show light drinkers outliving abstainers. That finding has been substantially undermined by two methodological problems:
- Sick quitters. The “abstainer” group in many cohorts contained people who stopped drinking because they were ill.
- Confounding by socioeconomic status. Moderate drinking in many populations tracks with income, education and health behaviours.
Mendelian randomisation studies — which use genetic variants affecting alcohol metabolism, and are far less vulnerable to both problems — have generally not reproduced a protective effect.
The J-curve was mostly a description of who drinks moderately, not of what moderate drinking does.
What this does not mean
It does not mean a glass of wine is comparable to smoking; the effect sizes are not close. It means the honest framing has shifted from “moderate drinking is good for you” to “less is better, and the shape of the risk curve has no flat part at the bottom.”
Where that leaves any individual is a judgement about a trade-off, which is a personal decision rather than a scientific one.