Healthy Versus Safe: What the Evidence Really Says About Alcohol

Every few months a new headline declares there is no safe amount of alcohol. If you enjoy a glass of wine with dinner, it is easy to either panic or bury your head in the sand. The missing detail that headlines miss is that there is a difference between a ‘healthy’ dose and a ‘safe’ dose.

For decades, cohort studies suggested light drinkers outlived non-drinkers. That J-shaped curve pointed to a well-known "a glass of red is good for your heart" message. It has not held up. A 2023 meta-analysis of 107 cohort studies and more than 4.8 million people found no reduction in all-cause mortality for people drinking under 25 g of alcohol a day compared with lifetime non-drinkers.

A big reason the old benefit disappeared is abstainer bias. Most earlier studies grouped former and occasional drinkers in with non-drinkers, and plenty of people stop drinking because they are unwell or have had a problem with alcohol. That makes the non-drinking group look worse off than it should.

The same analysis found something people quote less often. A significant increase in risk of death appeared at 25 g a day or more for women and 45 g a day or more for men. In New Zealand terms, that is roughly two and a half and four and a half standard drinks a day. So alcohol does not make you healthier, but the mortality data do not show measurable harm from the first sip either.

There is, however, nuance within this. And when we look at specific health issues, it can be seen. Of course, all-cause mortality is the bluntest measure we have, but it hides important detail. A large 2026 Burden of Proof analysis of 843 studies linked alcohol to higher risk of cancers of the breast, bowel, oesophagus, mouth, throat and liver, among others. Alcohol is a known carcinogen, therefore it should be unsurprising that we see increased risk of cancer at the lowest intakes.

It is important to understand that the difference between relative and absolute risk matters. Australian data from more than 225,000 adults aged 45 and over found alcohol-related cancer risk rose 19% for every additional ten drinks a week. That sounds alarming. In absolute terms, people drinking more than ten drinks a week had about a 4.9% higher cumulative risk of an alcohol-related cancer by age 85 than people drinking less than one a week. The risk is real, however it is less dramatic than a 19% headline implies. For someone having two or three drinks a week, the added risk is smaller again.

Pattern matters alongside total amount. In an earlier analysis of the same Australian cohort, breast cancer risk was marginally higher when a given amount of alcohol was concentrated into one to three days a week rather than spread across the week. A Spanish cohort of more than 18,000 adults found that a Mediterranean drinking pattern (moderate amounts, spread over the week, with meals, no binges) was linked with lower mortality at every level of intake. These are observational findings, and moderate drinkers tend to be moderate people, so we can’t rule out a lot of the lifestyle behaviours that place someone at reduced risk. Binge drinking, though, consistently looks worst.

Movement matters as well. A pooled analysis of eight British cohorts found that among people meeting physical activity guidelines, the link between alcohol and all-cause mortality was weakened and the link with cancer mortality almost disappeared. Exercise does not cancel out drinking, but it shifts the odds.

If you do not drink, there is no health reason to start. If you do, sensible guardrails look like this: no more than two drinks on any drinking day, more alcohol-free days than drinking days, drink with food, and skip the big nights. If cancer runs in your family, opting for lower limits of alcohol makes sense.

So why not just tell everyone to drink nothing? Because a message that does not match the data erodes trust, and all-or-nothing rules tend to collapse. When someone "fails" at total abstinence, the rest of the plan often goes with it. Most people get far more return from moving more, sleeping better and eating well than from shaving one drink off their week. And when sleep and stress improve, the evening drink often becomes less necessary on its own.

Four takeaways here:

1. Alcohol is not a health food. Once former and occasional drinkers are separated out, light drinking shows no mortality benefit over not drinking at all.
2. Healthy and safe are not the same thing. Mortality risk rises measurably above about two and a half drinks a day for women and four and a half for men, but cancer risk starts lower, so the practical ceiling should sit well below that.
3. Always ask for the absolute number. A 19% jump in relative risk translated to about 5% more cumulative cancer risk by age 85 for people drinking more than ten drinks a week. Context changes how scary a statistic is.
4. Pattern and lifestyle shift the odds. Spreading drinks out, having them with food, avoiding binges and meeting activity guidelines are all linked with lower risk.

Listen to more in the podcast episode.