The Intoxilyzer 8000 is the only breath instrument Florida approves as evidence, and the State treats its number as a fact. It is a measurement made by a machine, and like any measurement it rests on assumptions. This page walks through how the device works, because the places it can go wrong are the places a breath case is won.
As an ACS-CHAL Forensic Lawyer-Scientist and a member of the National College for DUI Defense (NCDD) National Task Force on alcohol and drug testing devices, I work with this science directly. Here is what the machine does with your breath, in plain terms.
Infrared Light, Not a Blood Draw
The 8000 never touches your blood. It works by infrared spectroscopy. A lamp sends infrared light through a captured sample of your breath, and alcohol molecules absorb specific wavelengths of that light. The more alcohol in the breath, the more light is absorbed. The machine measures the drop in light and converts it to a number. It is built by CMI, Inc. of Owensboro, Kentucky, and it is the only instrument Florida approves for evidentiary breath testing, so every breath case in this state runs through this one machine and its quirks.
Infrared light folds back and forth through a chamber of your breath. The more alcohol present, the more light it absorbs at the two wavelengths the detector watches. The machine turns that into an estimated blood-alcohol number.
Do not miss this
A breath reading at or above 0.08 also starts the 10-day license clock.
A breath result over the limit triggers an administrative suspension on top of the criminal case. You have 10 days from the arrest to demand a formal review hearing with the DHSMV in Clearwater, which protects your license and can secure a 42-day permit. We file that request the same day you hire us. Call or text (727) 761-4318.
The Folded Light Path and the Two Wavelengths
To read a thin sample accurately, the 8000 folds the infrared beam back and forth inside the sample chamber, which lengthens the distance the light travels through your breath. It watches two wavelengths, near 3.4 and 9.3 micrometers, where ethanol absorbs strongly. Watching two bands is meant to help the machine separate alcohol from other compounds. It does not always succeed, which is the heart of the interferents problem.
Two Samples and the Agreement Rule
Florida requires two separate breath samples, taken at least one minute apart, and the two results have to agree within a set tolerance. When they do not agree, the test is invalid. The printout records both samples, the air blanks, and the control checks the machine runs on itself. Reading that printout line by line is part of every breath case. See what a breath test printout means.
The criteria a sample has to meet
The machine is supposed to accept a sample only when it meets specific minimums, and those minimums are worth knowing, because they are checkable. An adequate breath sample has to run for at least 1 second, deliver a volume of at least 1.1 liters, and show an acceptable slope, meaning the alcohol reading climbs and then levels off into a plateau rather than spiking. The instrument also requires two adequate samples taken at least a minute apart, with self-checks at the start and end. Those rules exist because a short puff, a thin sample, or a sample that never plateaus is not the deep-lung air the result assumes. So when I read a printout, I am checking whether your samples cleared those bars or whether the machine bent its own rules to produce a number.
Why It Estimates Instead of Measures
Even working perfectly, the 8000 reports breath alcohol and then assumes a fixed ratio to estimate blood alcohol. That assumption, 2,100 to 1, is the same for everyone who blows into it, and real people vary. The machine also assumes a clean deep-lung sample, a quiet electronic environment, and an operator who followed every step.
Each assumption is a place to test. The ratio is the subject of the partition ratio. The clean-sample assumption is where mouth alcohol comes in. The machine’s own error and the room around it are covered in machine error and uncertainty.
An Aging Design
The 8000 runs on software and hardware designed in the early 2000s. It is the instrument Florida chose, and it carries the limits of its era. One detail tells the story: the manufacturer warrants the machine for a single year, warrants its software for only 90 days, and does not warrant it as fit for any particular purpose. A device the State treats as conclusive proof of guilt carries less of a guarantee than a household appliance. Knowing those limits, and how to show them to a jury, is what a forensic review brings to a breath case.
A valid test needs two samples that agree within 0.02, and the lower number is the one reported. That tolerance is the program’s own acknowledgment that the method carries spread.
The Number Inherits Its Assumptions
Once you see how the device works, the weak points follow naturally. The result is not a direct reading of blood. It is a breath measurement run through fixed assumptions, then printed as a blood-alcohol number. It assumes a blood-to-breath relationship that may not be yours, covered on the partition ratio page. It assumes a breath temperature it never measures, covered on the breath temperature page. And because it identifies alcohol by infrared absorption, other compounds can be read along with it, covered on the interferents page. The number on the card carries all of those assumptions inside it, even though it is presented as a simple fact.
People hear “the machine” and assume it measured something exact, and it estimated, on assumptions, under rules it does not always follow. I know this instrument, from the infrared method it uses to the minimum criteria each sample has to meet, and I read your printout against those standards line by line. When a sample fell short on time, volume, or slope and the result was reported anyway, that is not a clean test, and I make sure the jury understands the difference between a number and a measurement.
I started out as an Assistant Public Defender in Florida’s Thirteenth Judicial Circuit, in Tampa, and today I am one of six ACS-CHAL Forensic Lawyer-Scientists in Florida, with forensic training in how these instruments work. A breath result is an estimate produced by a machine, and I read its calibration history, its agreement checks, and the assumptions built into the number, so I can show a jury where it does not hold up. Learn more about my background.
Questions About the Intoxilyzer 8000
Does the Intoxilyzer 8000 test my blood?
No. It measures alcohol in a sample of your breath using infrared light, then estimates a blood-alcohol level from it using a fixed ratio. It never analyzes blood.
Why does the machine use two wavelengths?
Ethanol absorbs infrared light at specific wavelengths. The 8000 watches two of them, near 3.4 and 9.3 micrometers, which is meant to help separate alcohol from other compounds in the breath. It is not foolproof.
Why does Florida require two breath samples?
Two samples that agree within a set tolerance are a basic reliability check. If the two readings disagree, the test is not valid, and that is one of the things we look for on the printout.
Is the Intoxilyzer 8000 accurate?
It can be reliable when it is calibrated, maintained, and operated correctly. Its result still depends on the partition-ratio assumption, the testing environment, and the operator, any of which can move the number away from your true level.
What makes a breath sample valid on the Intoxilyzer 8000?
The machine requires two adequate samples taken at least a minute apart, and each must run at least 1 second, reach a volume of at least 1.1 liters, and show an acceptable slope that rises and then plateaus. A sample that falls short is supposed to be flagged, so whether your samples met the criteria is a checkable fact in the printout.
Related: the main breath test defense page, how we challenge a breath test, and the science behind the partition ratio.
This page is general information, not legal advice, and it does not create an attorney-client relationship. Breath testing in Florida is governed by Fla. Stat. 316.1932 and 316.1934 and the Florida Administrative Code chapter 11D-8. Procedures and rules change, and every case turns on its own facts. Past results do not guarantee a similar outcome.
