A California DUI test result measures one moment: the moment the sample was taken, at a station or a hospital, almost always well after you were last behind the wheel. The crime is defined by a different moment. Vehicle Code 23152(b) makes it unlawful to drive with a blood alcohol concentration of 0.08 percent or more, and the operative words are "to drive." The bridge California built between those two moments is the three-hour presumption, and it sits in a single sentence that almost nobody reads. What that sentence does, and what it cannot do, shapes how I defend every case that turns on how long it took to get you tested.

What the statute actually says

Vehicle Code 23152(b) is three paragraphs long. The first creates the offense. The second defines what "percent, by weight, of alcohol" means for blood and for breath. The third is the one at issue here:

Vehicle Code 23152(b), third paragraph. In any prosecution under this subdivision, it is a rebuttable presumption that the person had 0.08 percent or more, by weight, of alcohol in his or her blood at the time of driving the vehicle if the person had 0.08 percent or more, by weight, of alcohol in his or her blood at the time of the performance of a chemical test within three hours after the driving.

Read slowly, that sentence exposes the entire structure of a per se DUI case. Nobody can put a breath machine inside your car at the moment you were driving, so the Legislature gave the prosecution a way to reason backward from the test to the wheel. That is all this paragraph does. It is a rule about proof, not a second definition of the crime, and by its own terms it is written to be argued with.

The three hours run from the driving, not from the arrest

This is the most common misreading I hear, and it matters. The clock does not start when the patrol car lights come on, when you are handcuffed, or when you arrive at the station. It starts at the driving. It stops when the chemical test is performed, which for a blood case means the draw and for a breath case means the evidentiary blows at the station, not the handheld roadside PAS screening device and not the date the crime lab later analyzed the vial.

Practically, three hours is a long window, and most DUI arrests land inside it comfortably. A stop at 11:20 p.m. and a breath test at 12:05 a.m. is routine. But not every case is routine. Collision investigations, hospital transports, warrant applications for a blood draw after a refusal, and long backups at a busy county jail all push the test later. When the sample lands outside three hours, the presumption is simply unavailable, and that changes what the prosecution has to do.

There is a second timing problem hiding in the same sentence. The presumption is keyed to a test performed "within three hours after the driving," which means the prosecution has to establish when the driving happened. In a case built on an officer watching a car move, that is easy. In a case where the car was already parked when police arrived, where a citizen reported the driving hours earlier, or where the only evidence of when you drove is what you said to the officer, the time of driving can be genuinely uncertain. If the People cannot pin that time down, they cannot show the test fell inside the window, and the presumption never attaches at all.

It is an inference, not a shortcut around reasonable doubt

The statute uses the words "rebuttable presumption," and that phrase does real work in civil cases: it can shift a burden onto the other side. In a criminal case it cannot. A presumption that requires a jury to find an element of a crime against a defendant unless the defendant disproves it collides with the presumption of innocence, which is why the California Supreme Court held in People v. Roder (1983) 33 Cal.3d 491 that instructing a criminal jury in mandatory presumption language is unconstitutional. Evidence Code section 607 points the same direction: when a presumption operates in a criminal action to establish a fact essential to guilt, the facts giving rise to it must be found beyond a reasonable doubt, and the defendant need only raise a reasonable doubt about the presumed fact.

So the jury never hears the statute's wording. It hears CALCRIM 2111, the standard instruction for a 23152(b) count, which converts the presumption into a permissive inference. The instruction tells the jury that if the People have proved beyond a reasonable doubt that a sample of the defendant's blood or breath was taken within three hours of the driving, and that analysis of that sample showed 0.08 percent or more, the jury "may, but are not required to, conclude" that the blood alcohol level was 0.08 percent or more at the time of the alleged offense.

Three things follow from that phrasing, and they are the practical heart of this topic:

  • The People carry two predicate facts, beyond a reasonable doubt. That the sample was taken inside the three hours, and that the analysis showed 0.08 percent or more. Attack either one and the inference is off the table before the jury ever weighs it.
  • Even with both predicates proved, the jury is free to say no. "May, but are not required to" means exactly that. A juror who believes the timeline and the science point the other way is entitled to reject the inference outright.
  • The burden never moves to you. You are not required to prove your BAC was under 0.08 percent while driving. You have to raise a reasonable doubt, which is a far lower bar and is often reachable with the prosecution's own records.

What happens when the test falls outside three hours

A late test is not an inadmissible test. Nothing in Vehicle Code 23152 makes a result taken three and a half hours after driving disappear, and prosecutors still put those numbers in front of juries. What changes is that the shortcut is gone. Without the statutory inference, the People have to prove your blood alcohol at the time of driving some other way, and in practice that means calling a criminalist to perform retrograde extrapolation: estimating backward from the test result using assumed absorption and elimination rates.

That testimony is far more vulnerable than a bare number. Retrograde extrapolation only works if the expert can say where you were on the alcohol curve, and that depends on facts the state usually does not have: when you started drinking, when you stopped, what and how much, what you ate, and when. An expert forced to assume you were already past your peak has assumed the answer. This is the same ground the rising BAC defense occupies, and a test outside the three-hour window hands that defense a much larger opening.

How the presumption gets rebutted

Rebutting the inference is not one argument. It is a choice among several, and which one fits depends on your records:

  • Rising blood alcohol. If you were still absorbing your last drink when you were stopped, your level at the wheel was lower than the level the machine recorded later. The inference points backward in a straight line; absorption is not a straight line. See the rising BAC defense.
  • Drinking after driving. Alcohol consumed after you stopped driving is in the test result and was never in your blood behind the wheel. See the drink after drive defense.
  • The result itself is wrong. The second predicate fact is that the analysis showed 0.08 percent or more. Mouth alcohol, a broken observation period, a reflux condition, or a device with a bad maintenance history all go directly at that number. See mouth alcohol, GERD and acid reflux, and the calibration defense.
  • The timeline does not hold. The first predicate fact is the three-hour window itself. Unclear time of driving, a disputed collection time on the blood record, or a report whose times do not match the dispatch log can put the window in doubt.

Notice that the first two accept the test number and attack the inference drawn from it, while the third attacks the number and the fourth attacks the window. They are not mutually exclusive, and in a close case more than one is usually available.

A wrinkle worth knowing: breath results and partition ratio

Breath machines do not measure blood. They measure alcohol in breath and report a blood-equivalent figure using an assumed ratio, and real people vary from that assumption. You might expect that variation to be a standard way to attack a 0.08 percent reading, but the California Supreme Court held in People v. Bransford (1994) 8 Cal.4th 885 that partition ratio evidence is not admissible against the per se count, because the Legislature defined the offense in terms of breath alcohol as well as blood alcohol. Fifteen years later, in People v. McNeal (2009) 46 Cal.4th 1183, the court held that the same evidence is admissible on the impairment count under Vehicle Code 23152(a). So one body of science can be barred from the count the three-hour presumption serves and welcome on the count charged alongside it. That split is worth understanding before you assume a breath argument is available.

The same three-hour language appears in other subdivisions

The three-hour presumption is not unique to the standard 0.08 percent charge. The Legislature repeated it, with the applicable limit swapped in, in several places:

  • Vehicle Code 23152(d), driving a commercial motor vehicle at 0.04 percent or more, carries the identical presumption at the 0.04 percent figure. That matters a great deal to a commercial license holder, because the margin between a lawful level and a charge is half the size.
  • Vehicle Code 23152(e), driving a passenger-for-hire vehicle with a passenger aboard at 0.04 percent or more, carries it as well.
  • Vehicle Code 23153, DUI causing injury, carries the same presumption in its per se subdivisions at both the 0.08 percent and 0.04 percent thresholds.

Where the three-hour presumption does not apply

This is where most of the confusion I see comes from, because three different rules get blended into one.

It does not apply to the impairment count. The presumption is expressly limited to "any prosecution under this subdivision," meaning subdivision (b). The companion charge under Vehicle Code 23152(a) is governed by a different statute entirely, Vehicle Code 23610, which sets out its own presumptions: under 0.05 percent, a presumption that the person was not under the influence; between 0.05 and 0.08 percent, no presumption either way; 0.08 percent or more, a presumption that the person was under the influence. Section 23610 contains no three-hour clause. It speaks to what the chemical analysis shows about the person's condition at the time of the alleged offense, which is a different question from when the sample was collected.

It does not appear in the underage 0.05 percent statute. Vehicle Code 23140, which makes it unlawful for a driver under 21 to drive at 0.05 percent or more, contains no rebuttable presumption and no three-hour window. Neither does the zero tolerance provision. See underage DUI and zero tolerance.

It is not a DMV rule. Your Administrative Per Se hearing is a separate administrative proceeding, not a prosecution, and the suspension runs under Vehicle Code 13353.2, which contains no three-hour presumption. That does not mean timing arguments are pointless at the DMV. It means they are made with evidence rather than by invoking this statute, and the hearing has its own set of evidentiary presumptions about whether the test was properly administered. Treat the two forums as separate problems, because they are.

What I do with the timeline

Every one of the arguments above lives or dies on times, so the first thing I build in a close per se case is a minute-by-minute record. When did you last drink, and how do we prove it. What time does the report say you were driving, and what does the dispatch log say. What time was the arrest, the transport, the fifteen-minute observation period, each breath sample, or the blood draw. Do the times on the DS-367, the arrest report, the DMV discovery packet, and the lab paperwork agree with each other. They frequently do not, and each disagreement is either a hole in a predicate fact or an opening for a toxicologist. See my top DUI defenses for how this fits with everything else in a case.

Where your case sits on the clock

If your test was close to the three-hour line, or your reading was close to 0.08 percent, the timeline is worth a careful look before anything is decided. Use the free case analysis on this page, or call me directly at (888) 271-6644. I answer my own phone, 24/7.