Across those 42 people, peak BrAC (breath alcohol concentration) ranged from 0.040 to 0.117 g/210L - nearly a 3x spread from lowest to highest. More striking to me was the timing. Most people peaked within 13 minutes of stopping, but some were still climbing more than an hour after their last drink. A few had already peaked before finishing their final glass.

How it ran

Researchers recruited 42 adults for a supervised social drinking session. Everyone chose their own 80-proof mixed drinks - no fixed dose, no standardized schedule. Food was available throughout and eaten freely. Drinking sessions lasted anywhere from 45 to 172 minutes (mean 108, median 110), so some people were still going when others had been done for over an hour. BrAC was measured at approximately 30-minute intervals using a calibrated breathalyzer until each participant returned to zero.

That wasn't accidental. Most controlled alcohol studies fix the dose and cut out food to maximize pharmacokinetic precision. This one relaxed those constraints to see what variability looks like in something closer to actual social drinking.

The spread

The variation showed up across every measure:

  • Peak BrAC ranged from 0.040 to 0.117 g/210L (mean 0.087) - more than 2.5x from lowest to highest.
  • Timing was the stranger number. Time from last drink to peak ran from -23 to 76 minutes (mean 17, median 13). Most people crested within 13 minutes of stopping, but some were still climbing over an hour later. The negatives - peaks arriving before the final drink - aren't explained mechanistically in the abstract, though they probably reflect someone easing off sharply late in the session.
  • Elimination averaged 0.0186 g/210L/hr, ranging from 0.0134 to 0.0261 - nearly double between the fastest and slowest person.
  • A few participants showed multi-peak absorption: BrAC rose, dipped, then climbed again before the final descent. Once elimination started, it ran linear.

Where averages break down

I'd flag the study's limits first. Participants chose their own drinks and ate freely, so part of the peak variation reflects dose differences - someone who drank more will peak higher. Food intake wasn't quantified, which matters because food slows absorption sharply. The sample was 42 people, each measured only once, so there's no data on how reproducible any individual's profile is from night to night.

The timing data is harder to explain away. A 76-minute post-drink peak is real regardless of dose. So is the nearly 2x spread in elimination rates (0.0134 to 0.0261 g/210L/hr). Both point to real individual variation in how alcohol moves through the body.

What strikes me is how far this sits from the assumption built into most generic alcohol calculators: that you're an average metabolizer on an average night. Two people might have identical inputs - weight, sex, drinks consumed, time elapsed - and clear alcohol at rates that differ by nearly 2x. One might peak 76 minutes after stopping. The generic tool can't know that.

On a night you ate late, drank slowly, or just happen to clear alcohol slowly, the gap between a formula and your actual curve can be wide. And wide in the wrong direction: thinking you've peaked when you're still climbing.

That gap is what AlcoBalance is built around. Instead of a generic estimate, it tracks your actual curve - when your peak hits, when it fades. On a night your pace crept up without you noticing, you can see where you actually are on that curve and ease off while you're still steering the evening.

Source: Journal of Analytical Toxicology, DOI