The formula, and what it assumes
Widmark's formula is short enough to write on a receipt. Take the grams of pure ethanol consumed, divide by body weight in grams multiplied by a distribution factor, express as a percentage, then subtract a fixed amount per hour elapsed. The grams come from volume times strength times 0.7894, ethanol's density. The distribution factor represents the fraction of the body that alcohol actually distributes into, which is body water rather than the whole body — commonly quoted around 0.68 for men and 0.55 for women, though the original values and every subsequent revision are averages of small samples.
Every one of those steps assumes something that is not true for an individual. It assumes all the alcohol is already absorbed and evenly distributed, which takes 30 to 90 minutes after the last drink and longer with food. It assumes body water tracks weight and sex, when it actually tracks lean mass, so two people at 170 pounds with different body composition get materially different concentrations from the same drinks. It assumes elimination starts at the first drink and proceeds at a constant rate, when the rate varies roughly twofold across people and is altered by medication, liver condition and habitual intake.
This calculator makes no reduction for incomplete absorption. Some implementations multiply the grams by a factor below one, which lowers the output. Leaving it out biases the estimate high, which is the direction to be wrong in when the alternative is someone reading a low number as reassurance.
How large the error actually is
Validation studies that compare Widmark estimates against measured concentrations in the same people find scatter of tens of percent, with individual cases much further out. That is before you add the uncertainty in the inputs. Pour sizes at home are not standard servings; a generously poured glass of wine can be double the nominal five ounces. Beer strength runs from 4 to 9 percent across a single cooler. Timing gets remembered imprecisely, and "since the first drink" is a different quantity from "since the last drink" in a way the arithmetic is sensitive to.
Put together, an estimate from this page can sit far enough from a measured value that any threshold-based conclusion drawn from it is worthless. That is not a hedge in the small print, it is the central fact about the tool.
Standard drinks are a more useful number
The one figure here that survives all this is grams of ethanol, and its cousin the standard drink. In the United States a standard drink is 14 grams of pure alcohol, which is the amount in a 12 fluid ounce beer at 5 percent, a 5 ounce glass of wine at 12 percent, or a 1.5 ounce shot of 40 percent spirits by design. Those three are equivalent in alcohol and are the reason drink counts appear in public health guidance in the first place. Other countries define the unit differently — 10 grams in much of Europe and Australia, 8 grams in the UK — so drink counts do not translate directly across guidance documents.
Counting drinks is robust because it does not require any model of your body. Estimating a blood concentration does, and that is where the whole thing falls apart.
The part that is not arithmetic
Impairment of reaction time, judgement and attention begins at concentrations well below any legal threshold, and it is not something a person can assess from the inside — the same alcohol that impairs also impairs the ability to notice the impairment. Sleep does not accelerate clearance; it removes the sensation of intoxication while leaving the concentration on its normal downward slope, which is exactly the combination that produces morning driving with alcohol still present. If clearing time is the question, the sibling page for that is the alcohol elimination time calculator, and it carries the same caveats.
Legal limits are set by jurisdiction, differ between states and countries, and are lower for commercial drivers and for drivers under 21, with several places operating zero-tolerance rules. No number produced here maps onto any of them. Plan the ride before the first drink and the calculation never has to matter.
Questions people ask
Can I use this to work out whether I am under the limit?
No, and it deliberately prints no verdict. There are three separate reasons. The estimate itself carries individual error large enough to move it across any threshold you might compare it against. The inputs — pour size, strength, timing — are themselves approximate in ordinary use. And the threshold is not a single number: limits vary by state and country, and are set lower for commercial licence holders and for drivers under 21, with zero-tolerance rules in some places. A calculator that produced a pass or fail from those ingredients would be manufacturing false confidence. After drinking, the answer that carries no risk is not to drive.
Why does the same drink give a different number for different people?
Because alcohol distributes into body water, not into the whole body, and body water as a fraction of weight varies with lean mass. Two people at the same weight, one with more muscle and one with more fat, have different volumes of distribution and reach different concentrations from identical drinks. The formula handles this with a single factor keyed to sex, which is a crude population average and not a description of any individual. Elimination rate varies too, roughly twofold across people, and is affected by medication, liver function and habitual intake. That is why the page shows a range of clearing times rather than one.
Does coffee, food or a shower do anything?
Not to the concentration. Alcohol is metabolised in the liver at a rate that is essentially fixed for a given person and does not respond to caffeine, cold water, exercise or a large meal after the fact. What those things change is alertness, which is precisely the dangerous part: they can make someone feel considerably more capable than they are while the alcohol in their blood is unchanged. Eating before or during drinking does matter, but by slowing absorption and lowering the peak, not by clearing anything faster. Only time reduces the concentration.
How accurate is the elimination rate of 0.015 percent per hour?
It is a commonly cited middle value for a quantity that ranges roughly from 0.010 to 0.020 percent per hour across people, which is why the page shows all three. The metabolism is close to zero-order at ordinary concentrations, meaning it proceeds at a roughly constant absolute rate rather than proportionally to how much is present, so drinking more does not speed clearance — it just extends the time. Habitual heavy drinkers tend towards the faster end through enzyme induction, but not beyond that range. For planning purposes the slow figure is the one to take seriously.