How the estimate works.

1.The shape of the curve

Alcohol reaches your blood from the gut, rises to a peak, and then your liver removes it at a roughly steady pace. The model treats your body as a single compartment: alcohol flows in (absorption) and is cleared at a near-constant rate (elimination).

That gives the familiar shape, a quick rise followed by a long, nearly straight decline. It's Widmark's model with an absorption phase added, the standard description in forensic and clinical alcohol research.

Sources: [1][3]

2.Your body

How high the same drink takes you depends on how much body water the alcohol spreads into. Widmark used one fixed factor per sex. CheersCount uses the regression by Seidl and colleagues, which estimates it from sex, height and weight, and which matched their drinking experiments better than the alternatives they tested.

Age isn't part of that equation, so the app never asks for your birthday.

Sources: [1][2]

3.Absorption takes time

A drink isn't in your blood the moment you finish it. On an empty stomach the peak typically comes 30 to 45 minutes later; after a substantial meal it can take 90 to 120 minutes.

The model spreads every drink over its own absorption window, so a quick round and a slow evening look different even when the total is the same.

Sources: [3][8]

4.Drink strength

Stronger isn't simply faster. Alcohol is absorbed fastest from drinks of around 20-30% and more slowly at both ends: from beer because it's dilute and there's a lot of liquid, from neat spirits because a high concentration slows how fast the stomach empties.

The model's fastest point is 22%. For the same amount of alcohol, that changes the estimated peak noticeably.

Sources: [5][6]

5.Food

Food is the biggest factor you control. A meal keeps alcohol in the stomach longer, so it reaches the blood more slowly, and more of it is broken down before it gets there. Both lower the peak.

How long the stomach takes depends mainly on calories, which is why the app asks for a meal size or your own estimate. Caloric drinks count as well: a regular cola slows the drinks after it like a snack would, a diet one does nothing.

In the best measured case, ten men drinking straight after a breakfast of about 1,000 kcal, the peak was 36% lower than on an empty stomach. The model gives 31% for that setup. Studies also found food raises the elimination rate for a while; the model leaves that out, and a meal isn't a way to sober up faster.

Sources: [7][8][9]

6.Water

Water doesn't lower your blood alcohol. Your body holds around 40 litres of water, so a glass changes almost nothing, and the liver doesn't clear alcohol any faster when you drink more fluid. The model ignores water on purpose.

Alcohol does make you lose fluid, roughly 10 ml of urine per gram of alcohol in older experiments and less in newer ones, and the app tracks that on a separate bar that never touches the curve. Whether drinking water helps the morning after is less clear than people think.

Sources: [10][11]

7.Elimination and calibration

Your liver clears alcohol at a close to constant rate. The average for moderate drinkers is about 0.15 ‰/h, and people range widely around it. Unless you've measured, the app uses the average.

I deliberately don't let you type in a rate. A flattering guess wouldn't just shorten the tail, it would lower the whole curve.

If you own a breathalyzer, two readings on the way down, at least 20 minutes apart, give your own rate directly. Readings from several nights are pooled, pulled toward the average while there's little data, and kept between 0.08 ‰/h and 0.30 ‰/h.

Sources: [3][4]

8.Where it stops being reliable

The model describes a typical body. It knows nothing about these:

• Medication that slows the stomach (GLP-1 drugs, for example) or weight-loss surgery can shift the curve a lot, and the model doesn't account for either.
• The real peak can come an hour or more after the last drink, especially after a big meal.
• Close to zero, elimination slows down, which the model doesn't capture yet. Treat the time until zero as approximate.
• Zero on the curve isn't the same as recovered. Attention, memory and reaction time can stay impaired the next day with almost no alcohol left.
• It can't tell you whether you can drive, legally or safely. Don't use it for that decision.

Sources: [12]