Alcohol and health: the updated evidence after the ‘safe level’ claim collapsed

The “moderate drinking is fine” position has been significantly revised. Here is what the current evidence says and where the genuine uncertainty lies.

For about thirty years, the standard line was that moderate drinking was mildly good for you — a glass of red wine as cardiovascular maintenance. That position has been substantially dismantled, not by new moralism but by better methods. The J-shaped curve that underpinned it turned out to depend on who ends up in the “non-drinker” comparison group, and once that problem is addressed the curve largely disappears.

Where the J-curve came from, and why it collapsed

Dozens of cohort studies found that light-to-moderate drinkers had lower mortality than people who drank nothing. The inference was that moderate drinking conferred protection.

Also read

Supplement Reviews

WeightWorld Omega 3 review: the right numbers, minus the sourcing anyone can check

28 Aug 2026

Supplement Reviews

Ancient + Brave True Collagen review: the cleanest label in the category, the loosest claims

28 Aug 2026

The problem is the reference group. Lifelong abstainers are a minority of non-drinkers; the category also contains people who stopped because of illness, medication, addiction, or a doctor’s instruction. Comparing drinkers against a group enriched with the recently unwell will make drinking look protective. This is known as the sick quitter effect, and alongside it sits ordinary confounding — in most Western populations, moderate drinkers tend to be wealthier, better educated, more socially connected, and healthier in ways that are hard to fully adjust for.

A 2023 systematic review and meta-analysis pooling over a hundred studies and several million participants examined this directly. When analyses were restricted to higher-quality studies that avoided abstainer bias, the apparent mortality benefit of low-volume drinking did not survive.

The genetic evidence

The most persuasive line of argument comes from a design that sidesteps confounding entirely. Some people carry gene variants that make drinking physically unpleasant — flushing, nausea — and therefore drink far less throughout life. Because those variants are allocated at conception, they are not correlated with income, education, or health behaviour.

Studies using this approach, including a very large one in China, found that genetically lower alcohol intake was associated with lower blood pressure and lower stroke risk in a straight-line relationship. There was no sign of a protective sweet spot at low intake. If moderate drinking genuinely protected the heart, this method should have detected it. It did not.

The finding that matters most here

Alcohol is classified as a Group 1 carcinogen — the same category as tobacco and asbestos, which describes the strength of the evidence that it causes cancer in humans, not the size of the risk.

For women, the specific concern is breast cancer, and it is the reason the revision matters more for this readership than for men. Dose-response meta-analyses find breast cancer risk rising from low levels of intake, in the region of a 7 to 10% increase per additional drink per day, with no threshold below which no increase is detectable. The proposed mechanisms are acetaldehyde, a carcinogenic metabolite, and alcohol’s effect on circulating oestrogen.

Two physiological differences compound this. Women have proportionally less body water, so the same drink produces a higher blood alcohol concentration, and less of the gastric enzyme that begins breaking alcohol down before it reaches the bloodstream. Women also progress to alcohol-related liver disease at lower cumulative intakes than men.

What the guidelines now say

UK guidance sets a limit of 14 units a week for both men and women, spread across three or more days, and states explicitly that no level of drinking is completely safe. That guidance dropped the previously higher male limit precisely because the evidence no longer supported one.

Canadian guidance went further in 2023, reframing the question as a risk continuum: two drinks a week or fewer as low risk, three to six as moderate, and rising from there. US dietary guidelines remain more permissive at up to one drink daily for women.

The direction of travel is consistent across all of them, even where the numbers differ.

Where the genuine uncertainty lies

Being accurate here means resisting the temptation to overcorrect.

“No safe level” is a statement about relative risk having no detectable threshold, not a claim that a glass of wine carries meaningful individual danger. The absolute risk increase from light drinking is small in most contexts, and considerably smaller than the risks from smoking, obesity, or inactivity — which tend to get less alarmed coverage.

There is also a real limitation in what this research measures. Alcohol appears in most societies as part of eating, celebrating, and gathering, and the social value of those things does not appear in a mortality analysis. A framework that counts only physical harm will always conclude that zero is optimal, because zero is where the harm curve starts. That is a correct answer to a narrow question.

What has genuinely changed is the claim that drinking is beneficial. That claim no longer has support. Whether the small risks of light drinking are worth its non-medical value is a judgement the evidence cannot make for you.

The effect you can actually feel

Separate from the long-term risk picture, alcohol has a well-documented and immediate effect on sleep. It shortens the time taken to fall asleep, which is why it feels helpful, then suppresses REM sleep and fragments the second half of the night as it is metabolised. One or two drinks is enough to measure this.

For women in perimenopause, it compounds two symptoms already common in that period — disrupted sleep and hot flushes, which alcohol is a recognised trigger for. This is the part of the alcohol picture with the shortest feedback loop, and the easiest to test on yourself over a fortnight.

The practical takeaway

The honest summary is that drinking is not good for you, that light drinking is not very bad for you, and that the older advice suggesting it was actively protective was wrong. If you drink within guidelines and it adds something to your life, the evidence does not demand you stop — it removes the health justification for continuing, which is a different thing.

The decisions with the clearest supporting evidence are modest ones: fewer drinking days rather than smaller glasses on the same number of them, and moving drinks away from bedtime. And if cutting down proves harder than expected, or if you find yourself drinking to manage sleep or anxiety, that is worth a conversation with a GP — stopping heavy regular drinking abruptly can be medically risky, and doing it with support is safer than doing it alone.