Most BAC (blood alcohol concentration) math online leans on one tidy number for how fast the body clears alcohol. I've always been a little suspicious of that number, partly because I build an app that has to estimate it. A new paper in Alcohol, Clinical and Experimental Research measured elimination where it rarely gets measured: high concentrations, after long, real drinking sessions. The answer is messier than the calculators suggest.
Why hadn't anyone done this already? Most of what we know about breath alcohol elimination comes from low-dose studies. DUI (driving under the influence) cases, where these rates actually get used, usually involve much higher levels. So the reference values came from one range and got applied in another.
What they did
The researchers ran seven drinking sessions between October 2025 and February 2026, each lasting 11 to 13 hours. In total, 122 healthy volunteers took part, none of them alcohol dependent. For the elimination analysis they kept 91 people (43 women, 48 men) who gave samples during a confirmed elimination phase, the stretch after the peak when levels only go down.
They measured two things side by side:
- BAC from whole-blood samples, in grams of alcohol per kilogram of blood (g/kg)
- BrAC (breath alcohol concentration) with an evidential breath analyzer, the type whose readings count as legal evidence, in milligrams per liter of breath (mg/L)
Then the statistics. Linear regression looked for background characteristics that tracked with faster or slower elimination. To see how wide the spread of rates really is, they used bootstrapping, which means resampling the data over and over. Quantile regression then checked whether those extremes held up.
These were heavy sessions. Women peaked at a mean BAC of 1.54 g/kg, men at 1.71 g/kg. Mean breath peaks were 0.73 and 0.78 mg/L.
What they found
- Blood. Women cleared alcohol faster: a mean of 0.200 g/kg per hour (median 0.201). Men averaged 0.169 (median 0.159).
- Breath. Same direction. Women averaged 0.096 mg/L per hour (median 0.095), men 0.079 (median 0.083).
- Spread. This is the part that stuck with me. The middle 95% of women cleared blood alcohol at anywhere from 0.116 to 0.292 g/kg per hour. For men it ran from 0.092 to 0.256. On breath, women ranged from 0.034 to 0.139 mg/L per hour, men from 0.028 to 0.137.
Overall, the rates sat at or slightly above the upper end of previously reported values. For forensic work, the authors propose a low and a high reference rate: about 0.03 and 0.14 mg/L per hour on breath, which corresponds to 0.10 and 0.25 g/kg per hour in blood.
What it means
To make the spread feel real, I did some crude division. Take a man at the average peak of 1.71 g/kg and assume he clears alcohol at a steady rate all the way down (a simplification, and I'm not sure it holds near zero). At the median rate, that's about 11 hours. At the slow end of the range, almost 19. At the fast end, under 7. Same peak, roughly a 12-hour difference.
The sex difference is real in this data. It's just much smaller than the spread inside each sex. Between the average woman and the average man, the gap in blood elimination is about 0.03 g/kg per hour. Between the fastest and slowest men, it's about 0.16. So "women clear faster" is true on average and tells you very little about any one person.
This fits with something I wrote about earlier: the timing of your peak varies a lot from person to person. Turns out the way down varies just as much as the way up. And you can't feel your own rate either. At high BAC, people are the worst at judging how drunk they actually are, which is exactly the range this study covers.
One honest limit: these were healthy, non-dependent volunteers in organized sessions, so the numbers say less about people who drink every day.
This is the study I keep thinking about when I work on the time-to-sober estimate in AlcoBalance. The app tunes its curve to you, not averages, but I won't pretend that erases a spread like this. What it gives you is a look ahead. Around midnight you can see roughly when tonight's peak fades, put that next to tomorrow's 9 a.m. plan, and decide about the next drink while you're still steering the evening. It's a planning tool for your tomorrow. It doesn't tell you whether you're fit to drive.
The forensic researchers settled on a range instead of a single number. I think that's the right instinct for the rest of us too.
Source: Alcohol, Clinical and Experimental Research, DOI


