Is any amount of alcohol good for you?

Studies measure intake in grams of pure alcohol. A US standard drink holds 14 g, a UK unit 8 g, and the global burden study counts 10 g^1,5. Risk depends on the amount, so the numbers below matter.

Cancer. The International Agency for Research on Cancer classifies alcoholic beverages as carcinogenic to humans (Group 1)⁶. It is not only a heavy-drinker problem: about 4.1% of new cancers worldwide in 2020 were attributable to alcohol, and drinking under 20 g a day (about one and a half US drinks) accounted for roughly 14% of those cases⁷.

Heart and blood pressure. In 599,912 current drinkers, each extra 100 g a week went with higher risks of stroke, heart failure and fatal hypertensive disease, and a slightly lower risk of heart attack⁵. Gene variants are fixed at conception, so they are less affected by lifestyle differences between drinkers and non-drinkers. Variants that make people drink less went with lower blood pressure and stroke risk^3,4, and in one analysis lower coronary risk³. In a large Chinese study, genetically predicted intake had no significant effect on heart attacks⁴.

Liver. Cirrhosis (scarring of the liver) was not more likely in occasional drinkers but risk climbs steeply from about five drinks a day, and women face higher risk than men at the same intake⁸. If you already have fatty liver, see our fatty liver guide.

Sleep. At every dose studied, alcohol shortened the time to fall asleep, yet it disrupted the second half of the night and delayed the first period of REM sleep⁹.

Heart rhythm. In a randomised trial of 140 regular drinkers with atrial fibrillation (an irregular heartbeat), those who stopped drinking had fewer recurrences over six months (53% versus 73%)¹⁰.

Cutting back. In that trial, people assigned to abstain cut from about 17 to 2 drinks a week, so large reductions are achievable with support¹⁰. Track a typical week, set a weekly cap you can keep, and treat difficulty sticking to it as a health issue to raise with a clinician, not a failure of willpower. None of the studies above tested hangover or "liver support" supplements; for what one popular ingredient has and has not shown, see our dihydromyricetin article.

Global burden

The Global Burden of Disease analysis combined 694 data sources on drinking with a new meta-analysis of 592 studies covering 23 health outcomes. Alcohol was the seventh leading risk factor for deaths and disability worldwide in 2016 and the leading one at ages 15 to 49. Although the authors noted possible protective effects on some conditions, the intake that minimised overall risk was zero standard drinks a week (95% uncertainty interval 0.0 to 0.8)¹. These are modelled estimates across all ages, not trial results.

Dose thresholds

Wood and colleagues pooled individual data on 599,912 current drinkers without cardiovascular disease from 83 prospective studies in 19 high-income countries (40,310 deaths over 5.4 million person-years). All-cause mortality was lowest at about 100 g a week or less. Per extra 100 g a week, stroke risk was 14% higher (HR 1.14, 95% CI 1.10 to 1.17), heart failure 9% higher (1.09, 1.03 to 1.15) and fatal hypertensive disease 24% higher (1.24, 1.15 to 1.33), while heart attack was 6% lower (0.94, 0.91 to 0.97). Compared with drinking up to 100 g a week, life expectancy at age 40 was about 6 months shorter at 100 to 200 g, 1 to 2 years shorter at 200 to 350 g, and 4 to 5 years shorter above 350 g⁵. Everyone studied was a drinker, so the data cannot show whether zero beats 100 g.

The J-curve and its biases

Zhao and colleagues pooled 107 cohort studies (4,838,825 participants, 425,564 deaths). After adjusting for sampling variation, former-drinker bias and study quality, occasional drinkers (under 1.3 g a day) had no significantly lower mortality than lifetime non-drinkers (RR 0.96, 95% CI 0.86 to 1.06), nor did low-volume drinkers (1.3 to 24 g a day; RR 0.93, P = 0.07). Risk was not significantly higher at 25 to 44 g a day (RR 1.05) but clearly higher at 45 to 64 g (1.19) and 65 g or more (1.35). Female drinkers had a 22% higher risk than female lifetime non-drinkers (RR 1.22, P = 0.03)². The low-volume estimate still sits below 1, so a small benefit cannot be excluded.

Genetic evidence

Holmes and colleagues meta-analysed 56 studies (261,991 people; 20,259 coronary and 10,164 stroke events) using ADH1B rs1229984, a variant linked to drinking less. Carriers drank 17.2% fewer units a week and had lower blood pressure and lower odds of coronary heart disease (OR 0.90, 95% CI 0.84 to 0.96) and ischaemic stroke (0.83, 0.72 to 0.95); the benefit was the same in every drinking category, including light and moderate drinkers³. In China Kadoorie Biobank (512,715 adults followed about 10 years; 161,498 genotyped), men's reported intake showed a U-shape, but genetically predicted intake did not: stroke risk rose steadily (intracerebral haemorrhage RR 1.58 per 280 g a week, 95% CI 1.36 to 1.84; ischaemic stroke 1.27, 1.13 to 1.43), while heart attack was unchanged (0.96, 0.78 to 1.18)⁴. These methods assume the variants act only through alcohol, and the Chinese men drank mainly spirits, so results may not transfer.

Cancer

Rumgay and colleagues estimated that 741,300 new cancers in 2020 (4.1% of all cases) were attributable to alcohol, most often in the oesophagus (189,700), liver (154,700) and breast (98,300). Heavy drinking (over 60 g a day) accounted for 46.7%, risky drinking (20 to 60 g) 39.4% and moderate drinking (under 20 g) 13.9%, or 103,100 cases⁷. The figures are modelled, taking lifetime abstention as the zero-risk reference and a 10-year lag from drinking to diagnosis.

Liver, heart rhythm and pregnancy

A meta-analysis of 9 studies (2,629,272 participants, 5,505 cirrhosis cases) found no extra risk in occasional drinkers; one drink a day raised risk in women but not men; and five or more drinks a day raised it in both (women 5 to 6 drinks RR 12.44, 95% CI 6.65 to 23.27; men 3.80, 0.85 to 17.02)⁸. In the atrial fibrillation trial, 140 patients drinking 10 or more standard drinks a week were randomised to abstain or continue; recurrence was 53% versus 73% (HR 0.55, 95% CI 0.36 to 0.84), but it was open-label, lasted six months and enrolled mostly men drinking about 16 to 17 drinks a week¹⁰. In 34 cohort studies of prenatal exposure, binge drinking was linked to poorer child cognition (Cohen's d -0.13); mild-to-moderate exposure showed a tiny positive association that vanished when one large study was excluded, and the authors concluded no safe amount is known¹¹.

Where the evidence disagrees

Observational data still show lower heart-attack risk with more alcohol⁵, but the genetic studies do not confirm it^3,4, and the pooled mortality estimate for low-volume drinking leans below 1 without reaching significance². No trial has randomised people to years of light drinking, so the low-dose picture rests on observational and genetic methods.

Bottom line (confidence 4 of 5). Harm clearly rises with dose, and modelling attributes some cancers even to low intakes. The heart benefit of light drinking looks mostly like bias, though a small effect cannot be ruled out. If you do not drink, there is no health reason to start.