How much does alcohol shorten your life?
At age 40, the difference between the best and worst answer here is about 4.8 years of life expectancy — 5.0 for men and 4.6 for women. That is the gap between “1–7 a week” and “36+ a week”, holding everything else constant.
Alcohol is the factor where the science has moved most in the last decade, and the direction of travel has been consistently unkind. The old J-shaped curve — the idea that a moderate drink or two was protective — has largely collapsed under better methods.
What each answer costs, at 40
| Answer | Hazard ratio | Years vs. best |
|---|---|---|
| 1–7 a week | 0.99 | best |
| I don't drink | 1.00 | −0.1 |
| 8–14 a week | 1.05 | −0.6 |
| 15–21 a week | 1.15 | −1.5 |
| 22–35 a week | 1.30 | −2.8 |
| 36+ a week | 1.58 | −4.8 |
Computed from the SSA 2023 period life table with this factor’s hazard ratio applied and every other factor held neutral. Source: Wood et al., Lancet 2018 (599,912 drinkers).
Does it matter less as you get older?
Yes, and by a lot. A hazard ratio applies to the years you have left, so the same answer costs less in absolute terms the later you are asked it. That is an argument for changing it early, not for concluding it no longer matters.
| If you are now | Men | Women |
|---|---|---|
| 30 years old | 5.4 yrs | 4.8 yrs |
| 40 years old | 5.0 yrs | 4.6 yrs |
| 50 years old | 4.5 yrs | 4.2 yrs |
| 60 years old | 3.8 yrs | 3.7 yrs |
| 70 years old | 3.0 yrs | 3.1 yrs |
What the evidence actually shows
Wood and colleagues in the Lancet analysed 599,912 current drinkers across 83 studies and found no threshold below which more alcohol meant lower all-cause mortality. Above roughly 100g of pure alcohol a week — about seven standard drinks — life expectancy at 40 fell measurably, and the loss reached four to five years at the heaviest levels of consumption.
What to do about it
Risk climbs steadily above about 7 drinks a week. There is no dose that lengthens life, but the difference between 20 drinks and 7 is measured in years.
Where this number is weak
Much of the apparent benefit in older studies came from a classification error: lifelong abstainers were pooled with people who had stopped drinking because they were already ill, which made abstinence look dangerous. The model treats not drinking as neutral rather than harmful, which is what the corrected analyses support.
More generally: these are observational cohorts, which show association rather than proof of cause. The full methodology sets out every simplification the model makes.
What is it costing you?
The calculator weighs this against your other nine factors and ranks them.
Then make it count. The Extension Plan turns your result into a 12-week plan, starting with the habit worth the most years to you · $9. See what is in it →