Why does my estimated BAC differ from a breathalyzer reading?

A Widmark-based estimate is a mathematical model of average absorption and elimination; a breathalyzer measures actual breath alcohol at a moment in time. Food, medications, individual metabolism, and pour-size logging errors all move real BAC away from the modeled value, which is why estimates carry roughly ±20% variance. Neither an estimate nor a consumer breathalyzer should be used to decide whether driving is safe.

What elimination rate does the Widmark model assume?

A commonly used average is about 0.015% BAC per hour, but real elimination rates vary between roughly 0.010% and 0.020% per hour depending on the individual, liver function, and other factors. This variation is one of the main sources of uncertainty in any BAC estimate.

← Back to Home

The Widmark Formula Explained

Q: Is the Widmark formula still used today?

Yes. Nearly a century after Erik Widmark's work in the 1930s, the formula remains the foundation of forensic BAC estimation and is still referenced in toxicology, though experts apply it with documented uncertainty ranges. Modern refinements like the Watson Total Body Water method improve the body-composition side of the calculation, but the core model is unchanged.

Q: What is the Watson Total Body Water method?

Watson's 1980 equations estimate a person's total body water from height, weight, age, and biological sex. Because alcohol distributes into body water, deriving the Widmark distribution factor from estimated TBW — rather than using Widmark's fixed averages of about 0.68 for men and 0.55 for women — personalizes the estimate to an individual's body rather than a population average.

The math behind BAC estimation — the classic equation, the Watson refinement, a worked example, and where the model's limits are.

Last updated: July 3, 2026

Where the formula comes from

In the 1930s, Swedish physician Erik M. P. Widmark published the pharmacokinetic model that still underpins most blood alcohol estimation today. His insight was that BAC can be approximated from four things: how much pure alcohol was consumed, how much of the body it distributes into, and how much time has passed at what elimination rate.

The equation, in plain terms

The classic Widmark formula is:

BAC% = ( A / (r × W) ) × 100 − (β × t)

A — grams of pure alcohol consumed (a US standard drink is about 14 g; see our standard drink guide)
W — body weight in grams
r — the Widmark distribution factor: the fraction of the body alcohol effectively distributes into. Widmark's population averages were about 0.68 for men and 0.55 for women
β — the elimination rate, commonly averaged at about 0.015% BAC per hour (individual range roughly 0.010–0.020%)
t — hours since drinking began

Intuitively: alcohol spreads through your body water (the first term), while your liver removes it at a roughly constant hourly rate (the second term).

The Watson refinement: personalizing r

The weakest part of the classic formula is the fixed r: two people of the same weight can have very different body water. In 1980, Watson and colleagues published equations that estimate Total Body Water (TBW) from height, weight, age, and biological sex. Deriving the distribution factor from estimated TBW instead of a fixed average makes the estimate reflect your body rather than a 1930s population mean.

This is the approach SipLogger uses: Watson TBW to personalize the distribution factor, then Widmark's model for accumulation and elimination — with each drink's absorption modeled individually over time rather than assuming everything is absorbed instantly.

A worked example

Take a 70 kg (154 lb) man who has two US standard drinks (about 28 g of pure alcohol) and measure one hour after starting:

Distribution: 28 g ÷ (0.68 × 70,000 g) × 100 ≈ 0.059% peak estimated BAC
Elimination: after 1 hour at 0.015%/hour, subtract 0.015 → ≈ 0.044% estimated BAC

The same drinks for a 60 kg woman (r ≈ 0.55): 28 ÷ (0.55 × 60,000) × 100 ≈ 0.085% peak — nearly half again higher, from identical drinks. Body parameters matter enormously, which is why generic "drinks per hour" rules mislead.

Remember what these numbers are: modeled averages with roughly ±20% variance, not measurements. Food in the stomach, medications, and individual metabolism can move real values well outside the example figures.

What the model can't do

The Widmark model — even Watson-refined — assumes average absorption, a constant elimination rate, and accurate logging. In reality absorption varies with food and drink type, elimination varies by individual and session, and nobody logs a heavy pour perfectly. That is why any honest implementation, SipLogger included, presents results as educational estimates only — useful for understanding the shape of alcohol metabolism, never for deciding whether it is safe to drive. For the full picture, read how accurate BAC calculators really are.

Frequently Asked Questions

Is the Widmark formula still used today?

Yes. It remains the foundation of forensic BAC estimation and toxicology teaching, applied with documented uncertainty ranges. Modern refinements like Watson TBW improve the body-composition input, but the core model is unchanged after nearly a century.

What is the Watson Total Body Water method?

Watson's 1980 equations estimate total body water from height, weight, age, and biological sex. Because alcohol distributes into body water, deriving the distribution factor from TBW personalizes the estimate instead of using Widmark's fixed 0.68/0.55 averages.

Why does my estimated BAC differ from a breathalyzer?

An estimate is a mathematical model of averages; a breathalyzer measures actual breath alcohol at a moment. Food, medications, individual metabolism, and logging error all move real BAC away from the model — the source of the roughly ±20% variance. Neither should be used to decide whether driving is safe.

What elimination rate does the model assume?

A common average is 0.015% BAC per hour, with real individual rates ranging roughly 0.010–0.020%. That spread alone creates meaningful uncertainty in any estimate over a multi-hour session.