At Two Drinks a Day, 1 in 25 Men Die of Alcohol, a Federally Commissioned Study Estimates. By Our Math, About Triple the Average Man's Odds of Dying in a Crash

Among men who average two drinks a day for life, a federally commissioned study estimates that about 1 in 25 die of something alcohol caused. Two a day, which the study treats as 14 a week, was the daily ceiling the Dietary Guidelines for Americans set for men until January, when the 2025–2030 edition dropped its numeric alcohol limits and told Americans to “consume less alcohol for better overall health.” Sinead George, Kevin Shield and colleagues published the study in June in the Journal of Studies on Alcohol and Drugs, and it is back in the news. By our arithmetic, 1 in 25 is about three times a man’s lifetime odds of dying in a motor-vehicle crash.

How the Study Attributes Deaths to Alcohol

This study followed no one to the grave. Its authors reviewed more than 7,200 articles to fix a relative risk for each condition alcohol causes, then applied those risks to national drinking surveys, Census counts and CDC death records. Out comes a lifetime risk of an alcohol-attributable death at each weekly intake, summed over cancers, liver disease, cardiovascular disease and injuries: a population average, not a personal forecast. A federal body, the Interagency Coordinating Committee on the Prevention of Underage Drinking, commissioned the study to inform the guidelines. An editorial in the same issue says its dissemination stalled.

According to the study’s abstract, the risk tops 1 in 1,000 above about 6.5 drinks a week for men and 7 for women, and 1 in 100 above about 8.5 for both. Those lines are loose: the 95 percent interval on the men’s 1-in-1,000 crossing runs from under one drink a week to 13.5. For men at 14 a week, the paper’s table gives 39.3 per 1,000. George and colleagues recommend no more than one drink a day for both sexes.

About Three Times a Man’s Crash-Death Odds

We borrow the National Safety Council’s method. Take men’s 72 percent share of crash deaths in the Insurance Institute for Highway Safety’s 2023 figures, apply it to NHTSA’s early count of 36,640 traffic deaths in 2025, divide by the Census Bureau’s 168.3 million males (via Statista) and multiply by male life expectancy, 76.5 years: about 1 in 83. On the council’s count of 42,789 motor-vehicle deaths in 2024, which includes crashes off public roads, the odds are about 1 in 71. Against that band, which averages all men from birth, 39.3 per 1,000 is 2.8 to 3.3 times the risk. Call it three. The two overlap: alcohol-involved crash deaths count in both.

Across the study’s 95 percent interval, 9.6 to 69.6 per 1,000, the risk at the old ceiling runs from about three-quarters of a man’s lifetime crash-death odds to five to six times them.

What the Second Drink Costs

A micromort, a one-in-a-million chance of death, lets us cut a lifetime risk into single drinks. Going from 6.5 drinks a week, where men cross 1 in 1,000, to 14 adds about a drink a day and lifts the risk to 39.3 per 1,000, which over 50 drinking years means 19,600 extra drinks at two micromorts apiece. At NHTSA’s 2025 rate of 1.10 road deaths of any kind per 100 million vehicle miles, one micromort is about 91 vehicle miles, so each added drink carries, on average, as much death risk as about 180 vehicle miles of US traffic. A 40- or 60-year drinking life gives 220 or 150 miles; the study’s interval, at 50 years, gives 40 to 320. On the central estimate, though its interval is wide, drinks below 6.5 a week average under a tenth of a micromort each.

What You Can Do

Count in standard drinks of 14 grams of alcohol: a 12-ounce beer at 5 percent, a 5-ounce glass of wine at 12 percent or a 1.5-ounce shot of spirits. For a man drinking two a day, our arithmetic on the study’s central estimates puts over seven-tenths of the risk in the second drink, and probably more than nine-tenths. On central estimates, one a day, the old ceiling for women, sits at about 1 in 1,000 for women and between 1 in 1,000 and 1 in 100 for men. The model compares lifelong habits; it does not measure what cutting back midlife gains. Nothing here is a reason to start drinking, and age under 21, pregnancy, some medications or recovery from alcohol use disorder call for none at all. Trouble cutting back is treatable; SAMHSA’s free, confidential, 24-hour National Helpline is 1-800-662-4357.

Limitations

Every figure and quotation comes from search summaries; we could open none of the linked pages. Our per-drink figures assume a drinking life of 40 to 60 years because we could not see the paper’s age window, which may cover fewer years than our crash odds. Our road benchmark applies a 2023 male share and the 2024 male population to other years’ totals, holds today’s rates fixed, counts from birth and averages over all men, drivers or not, while the alcohol figure assumes a steady 14 drinks a week for life, from surveys that undercount drinking. Our miles count everyone a crash kills, not only drivers; crashes kill younger, and we compare deaths, not years lost. Nor could we confirm whether 39.3 nets out deaths the model says alcohol prevents. We contacted none of the authors or bodies named.

The Strongest Case Against This Analysis

A separate review, mandated by Congress, points the other way. In December 2024 the National Academies found with moderate certainty that moderate drinking was associated with lower all-cause mortality than never drinking, and defined moderate as up to two drinks a day for men, the very level in our headline. It excluded studies that lumped never-drinkers with former drinkers, who may have quit because they were ill. It also linked moderate drinking to higher breast-cancer risk in women. In January 2026 House Oversight Chairman James Comer released a staff report, “A Study Fraught with Bias,” arguing that the study duplicated work Congress assigned the Academies, was run in a way inconsistent with federal law and drew on a review panel whose members and process showed bias.

Those are real objections. SAMHSA’s study page describes that panel as external scientific experts, overseen by a technical review subcommittee; we could not test the bias charge. The reviews disagree on overall mortality but answer different questions. Both rest on observational data. The Academies compared death rates between groups, which, as STAT reported, does not show alcohol caused the difference. This study assumes each condition’s observational relative risk is causal, an assumption its method cannot test, and sums the deaths. Its small protective associations, at up to three drinks a week, missed statistical significance. If the Academies are right, two a day need not shorten men’s lives on net, even if 1 in 25 die of alcohol.

The Bottom Line

Federal guidance no longer names a number; a contested study commissioned to inform it does. At the old ceiling for men it is 1 in 25. By our arithmetic, that is about three times a man’s odds of dying in a crash, and on central estimates most of it sits in the second daily drink: over 50 drinking years, each of those second drinks carries about the death risk of 180 vehicle miles of US traffic. Its authors advise one a day at most.

Related

• GLP-1 Drugs Have Positive Evidence Against 7 of the 10 Leading Causes of Death in America. Only 4 Have Randomized Trial Support.
• We Ran the Numbers on 7 Energy Drinks. The Best One for Deep Work Costs 1.15¢ Per Milligram.