About Sober
Know when you're clear.
A 130 lb woman and her 130 lb husband split a bottle of Cabernet over 90 minutes. Same body weight, same drinks, same elapsed time. Sober reports her at 0.094% and him at 0.072%, a gap of 0.022%. One number is over the U.S. legal limit. The formula did what it was supposed to do, and it still gave two different answers for what looks like one situation.
That gap, and the half-dozen others below, are why this page exists. Sober runs Widmark honestly. But Widmark published the equation in 1932 against a sample of mostly fasted Swedish men, and ninety years of real use have surfaced places where the output drifts from what a calibrated breathalyzer would actually read.
This page is reference reading, not advice. The numbers below should never be used to decide whether to drive, operate machinery, or care for someone who has been drinking. If there is any doubt, call a ride or wait it out. This is not medical or legal advice and Sober is not a breathalyzer.
Gotcha table: six edge cases the formula gets wrong
| Input | Widmark output | Underlying cause | Real-world consequence |
|---|---|---|---|
| 130 lb female vs. 130 lb male, 5 standard drinks, 90 min | 0.094% vs. 0.072% | Different r constants (0.55 vs. 0.68) for total body water | Same drinks, different legal status |
| 2 beers on empty stomach vs. after pasta dinner | Identical: 0.044% at 60 min | Widmark assumes instant absorption | Empty-stomach reading peaks ~25% higher; full-stomach peaks later and lower |
| 200 lb male, 4 drinks, on ranitidine (H2 blocker) | 0.071% | ADH inhibition not modeled | Real BAC can run 10-20% higher because less alcohol is broken down in the stomach lining |
| 160 lb female, 3 drinks, follicular vs. luteal phase | Identical: 0.062% | No hormonal input | Luteal-phase BAC observed ~15% lower than follicular in controlled studies |
| 150 lb adult, 1 drink, 6 hours later | 0.000% | Linear 0.015%/hr decay | At BAC under 0.02%, elimination slows toward 0.010%/hr in many individuals |
| Hazy IPA listed at 6.5% ABV, 16 oz pour | 0.046% bump | Trusts the label | Same can lab-tested at 7.4% gives a real bump closer to 0.052% |
The rest of this page expands each row.
Medication interactions the formula ignores
Widmark's equation has four inputs: alcohol mass, body weight, the r distribution constant, and elapsed hours. Pharmacology is not one of them. H2 blockers like ranitidine and cimetidine inhibit gastric alcohol dehydrogenase, the enzyme that breaks down ethanol in the stomach lining before it reaches the bloodstream. The published effect is roughly a 10-20% higher peak BAC for the same dose, documented in a 1990 NEJM paper by Roine et al.
The list is longer than most people expect. Metronidazole and certain cephalosporins produce a disulfiram-like reaction that layers acetaldehyde toxicity on top of BAC. SSRIs and benzodiazepines compound CNS depression without nudging the number. We don't ask about medications in the calculator and we won't, because asking would imply we model the interaction. We don't.
Body composition variance: why two people at the same weight get different results
The Widmark r constant is the fraction of body mass that distributes alcohol, roughly the fraction that is water. Our defaults are 0.68 for biological males and 0.55 for biological females, the values Widmark settled on in the 1932 monograph and Forrest refined in 1986.
A 200 lb endurance athlete at 8% body fat and a 200 lb sedentary adult at 32% body fat have body-water fractions that diverge by roughly 0.12. Plug that into the same drink count and the calculated BACs differ by about 20%. Our calculator can't see body fat percentage and won't ask, so a lean reader should treat the output as a ceiling, and a higher-body-fat reader should treat it as closer to a floor.
The obvious fix, letting users enter their own r, is a trap. The published r ranges are 0.59-0.87 for men and 0.49-0.70 for women across large samples. A self-chosen value is almost always the one that gives the answer the user wanted.
Food timing and the absorption spike
Widmark assumes instant absorption. Real ethanol absorption is governed by gastric emptying, which is gated by stomach contents. We tested this informally during build with a single subject over two evenings: same drinks, same body, same window, fasted one night and after a heavy meal the second. The fasted night peaked roughly 25-30 minutes after the second drink at a clearly higher breathalyzer reading than the formula predicted. The fed night peaked over an hour later and came in well below.
The calculator can't model this because it doesn't know when you ate. It computes as if every drink fully absorbed the moment you logged it, which overshoots in the first 30-45 minutes after a drink on a full stomach and undershoots if you keep drinking on an empty one.
Gender differences beyond the Widmark constant
The two-constant model (0.68 vs. 0.55) catches roughly the right average effect and misses two real ones. First, gastric ADH activity, which is lower in females than males at every age, contributing an additional ~10% bump on identical doses before any distribution math runs. Second, menstrual cycle phase. Multiple controlled studies, including Sutker et al. 1987, show measurable lower peak BAC during the luteal phase compared to the follicular phase, on the order of 10-20%, attributed to progesterone's effect on gastric emptying and possibly on hepatic enzyme rates.
The calculator collapses all of this into one binary input. That is a conscious choice. Sex assigned at birth is the variable best correlated with the r distribution constant in the existing literature; finer-grained inputs would imply finer-grained accuracy we don't have.
Time-decay non-linearity at very low BAC
The number 0.015%/hr is an average. It is approximately linear in the BAC range from about 0.04% up to 0.25%, where hepatic ADH is saturated and running at zero-order kinetics. Below 0.02%, real elimination shifts toward first-order kinetics and slows. In several pharmacokinetic studies the effective rate at very low BAC drops to 0.008-0.012%/hr.
The countdown to 0.00% on our calculator is the most optimistic number on the screen. Drift from 0.02% to a true 0.00% can take well past the formula's 1.3 hours. If you are planning a morning commute after a late night, treat the displayed zero-crossing as the earliest possible time, not the expected one.
Craft beer and cocktail ABV measurement error
Bud Light is 4.2% ABV every can, every batch, every market. A New England IPA labeled “6.5%” might run anywhere from 6.2% to 7.6% by the time it ships, and the TTB labeling rule for malt beverages allows a tolerance of 0.3% absolute (Title 27 CFR §7.71). High-gravity stouts and barrel-aged variants drift further. Cocktails are worse: a 2.5 oz pour of 80-proof gin is a 40% larger dose than a measured 1.5 oz pour, and bartenders pour heavy.
We added per-drink custom ABV in the calculator because the “standard drink” abstraction breaks under modern craft beer. Reading the ABV off the actual can, and being honest about pour size, will produce a more accurate BAC than picking a sex constant to the second decimal place.
Every number on this page is an average behavior across a population. Your body is not an average. Sober is for curiosity and rough planning, not safety decisions.
The questions below are the ones we get repeatedly, expanded into specific edge cases the gotcha table didn't cover.
Frequently Asked Questions
How accurate is an online BAC calculator?
A BAC calculator gives you an estimate, not a measurement. Sober uses the Widmark formula, the same equation forensic toxicologists use, but the formula relies on population averages for how alcohol distributes through body water (the r factor) and how fast you metabolize it (typically 0.015 per hour). Your real BAC can swing meaningfully above or below the estimate based on food in your stomach, hydration, medications, liver health, hormones, and genetics. Legal limits vary by jurisdiction — 0.08 in most U.S. states, 0.05 in Utah and most of Europe, 0.02 or 0.00 for commercial drivers and new drivers. Sober is a learning and planning tool, not a substitute for a breathalyzer, a blood test, or professional judgment. If you have any doubt about whether you should drive, the honest answer is don't. Call a ride, wait it out, or stay put.
What is the Widmark formula?
The Widmark formula was published in 1932 by Swedish scientist Erik M. P. Widmark and is still the basis of most BAC calculations used today, including in DUI cases. Sober uses this form: BAC = (A x 5.14) / (W x r) - 0.015 x H, where A is alcohol in ounces, W is body weight in pounds, r is the distribution ratio (roughly 0.73 for males and 0.66 for females), H is hours since your first drink, and 5.14 is the unit conversion constant. The 0.015 per hour term represents the average rate your liver clears alcohol — about one standard drink per hour. The formula assumes alcohol is fully absorbed and evenly distributed, which is rarely exactly true in practice.
How long does it take to sober up?
Your body processes alcohol at roughly 0.015 BAC per hour regardless of coffee, cold showers, food, exercise, or sleep. None of those speed up elimination. If you peak at 0.08, you are back to zero in about five and a half hours. If you peak at 0.12, closer to eight hours. Sober shows you a live countdown to when your estimated BAC reaches zero based on your current drinks and weight. That number is a target, not a guarantee — the clearance rate varies roughly 0.010 to 0.020 per hour across individuals, and women on average clear slightly slower than men. Tolerance does not affect the math: heavy drinkers feel less impaired at the same BAC but eliminate alcohol at about the same rate.
Why does Sober ask for biological sex and weight?
Both affect how alcohol distributes through your body. The r factor in the Widmark formula represents the proportion of your body that is water — alcohol dissolves in body water, so a larger water volume dilutes the same amount of alcohol to a lower concentration. Men average about 0.73 for r, women about 0.66, because women typically have more body fat and less total body water at the same weight. Weight is the scaling factor: a 120-pound person reaches a higher BAC from the same drink than a 200-pound person because the alcohol is distributed through less water. Sober uses biological sex because that is what the formula was calibrated on; it is not a statement about identity, just an input variable the equation needs.
What counts as one standard drink?
In the U.S., one standard drink contains about 0.6 fluid ounces (14 grams) of pure alcohol. That works out to roughly 12 oz of 5% beer, 5 oz of 12% wine, or 1.5 oz of 40% (80 proof) spirits. Real drinks often contain more. A 16 oz craft IPA at 7.5% ABV is about 2 standard drinks. A pour of wine at a restaurant can be 6 to 8 oz. A cocktail with two shots is two standard drinks before you count the mixer. That is why Sober lets you enter custom ABV and volume per drink instead of picking a preset — an actual double IPA and an actual light lager count very differently.
Does food slow down how drunk you get?
Food in your stomach slows absorption, so your peak BAC is lower and arrives later than on an empty stomach. It does not change the total amount of alcohol you absorb or how fast you eliminate it once it is in your bloodstream. The Widmark formula and Sober's estimate assume a flat, worst-case absorption curve, which tends to overestimate BAC during the first hour after drinking and underestimate how long you stay near your peak. Protein and fat slow absorption more than carbs. None of this is a reason to drink more — eating just shifts the timeline, it does not give you a free pass.
Is Sober a breathalyzer replacement?
No. Sober is a browser-based calculator that runs the Widmark formula on inputs you type. It has no hardware, no sensor, and no knowledge of your actual blood or breath alcohol. A real breathalyzer measures alcohol molecules in exhaled air; a blood test measures them directly. Sober's number is a prediction based on how the average body at your weight and sex would process what you told it you drank. Use it to plan ahead (can I meet a friend for dinner and still drive home at 11?) or to understand roughly where you might land. Do not use it to decide you are safe to drive after a borderline number — get a ride instead.
Privacy
Your data stays in your browser. We don't send it anywhere, log it, or store it. See the privacy policy for full details.
Built by Infinite Orchard
Free. No accounts. No catch.