Ask Deandra: How Accurate Is the Breathalyzer in Texas?

Here is the longer answer: what the Intoxilyzer 9000 actually measures, where the errors come from, and how a forensic defense takes the number apart.

The Device: The Intoxilyzer 9000

The breath testing instrument in use across Texas is the Intoxilyzer 9000, manufactured by CMI, Inc. It replaced the older Intoxilyzer 5000 as the official breath alcohol instrument under the Texas Department of Public Safety Breath Alcohol Program. One point is worth being precise about, because it is widely misunderstood: whether to run a breath testing program at all is up to each individual police agency, not the county. Some agencies have shut their breath programs down and offer only blood testing to a person arrested for DWI. Where breath evidence is collected in Texas, though, the instrument is the 9000, in Dallas, Tarrant, Collin, Denton, McLennan, and Rockwall Counties and everywhere else our firm practices.

The 9000 is not a new invention. It is an iterative refinement of a decades-old measurement technology, with all the strengths and limitations of that technology. Understanding those limitations is the first step in understanding what a breath test result actually means.

The Roadside Device Is Not the Machine in Question

Before going further it is worth separating two things that both get called a breath test. The handheld unit an officer may offer you at the side of the road is a portable breath test, a screening tool the Texas Department of Public Safety has never certified as producing a reliable, accurate breath alcohol concentration. Its number is not admissible at trial.

That rule has a practical edge worth knowing. If the roadside reading was said out loud on the dashcam video, or held up to the camera, the number can still reach a jury unless someone moves to exclude it before the video is played. None of that happens on its own.

Everything below this point is about the other device: the fixed evidentiary instrument at a station or a jail, which has to be plugged in, warmed up, and maintained under a state program, and which is the only one that produces a number a prosecutor can put in front of a jury.

How the Intoxilyzer 9000 Actually Works

The 9000 measures alcohol indirectly, using infrared (IR) spectroscopy. When you blow into the instrument, a sample of deep-lung (alveolar) air is captured in a chamber. Infrared light is passed through the chamber, and the instrument measures how much of that light is absorbed at specific wavelengths associated with ethanol. The more absorption, the more ethanol in the breath sample.

The physical arrangement inside is simple to picture. A heated chamber the breath passes through, an infrared source at one end, a detector at the other. The source sends energy through the sample, the detector measures how much arrives, and the difference between the two is the energy something in the sample absorbed. The instrument reads at several wavelengths rather than one, because ethanol absorbs infrared energy in a characteristic pattern across several points, and comparing absorption across them is how the machine tries to tell ethanol apart from everything else that also absorbs infrared.

That is where the trouble starts. The machine does not measure blood alcohol. It measures breath alcohol, and everything after that rests on an assumption about the relationship between the two. What unit the resulting number is actually reported in matters more than most people expect, so it is worth pinning down before going any further.

The Test Is Two Samples, Not One

A Texas breath test is not a single blow. The sequence is automated and runs in a fixed order: the operator confirms the observation period was completed, the instrument runs an air blank to confirm the chamber is clean, it analyzes a reference standard of known alcohol concentration to check that it is reading correctly, another air blank, your first breath sample, another air blank, your second breath sample, and a final air blank.

Two samples are taken because a single measurement tells you nothing about its own reliability. Duplicate testing is ordinary forensic practice. If two samples taken minutes apart agree closely, that agreement is itself evidence the measurement was sound. If they disagree, something about the sampling deserves a second look.

Texas sets the tolerance at 0.02. The two results have to fall within that of each other for the test to count, and it is worth sitting with what that permits. A first sample of 0.09 and a second of 0.07 are treated as agreement. One of those numbers is over the limit and one is under it, and the test is still valid.

What Unit the Number Is Actually In

The figure on a Texas breath slip is not a blood alcohol measurement and does not claim to be. It is reported as grams of alcohol per 210 liters of breath. That is the unit Texas Penal Code 49.01 uses to define the per se offense in breath cases, alongside a separate blood standard expressed as grams per 100 milliliters of blood. A reported 0.09 means 0.09 grams of alcohol per 210 liters of breath, and the only thing tying that figure to a blood concentration is the assumption described next.

The Partition Ratio Problem

The Intoxilyzer 9000 assumes that, for every unit of alcohol in your breath, there are 2,100 units of alcohol in your blood. That 2,100:1 figure is called the partition ratio, and it is the single biggest source of forensic error in breath testing.

Real partition ratios vary from roughly 1,100:1 to 3,400:1 across the population. They vary by sex, by body temperature, by hematocrit, by breathing pattern, by the phase of alcohol absorption the person is in, and by dozens of other physiological variables. A person with a true partition ratio of 1,500:1 who is blowing a 0.09 on the 9000 actually has a blood alcohol concentration well below 0.08 because the machine is assuming a ratio that does not match that person’s physiology.

The fixed 2,100:1 ratio makes prosecution easier. It does not make the number an accurate estimate of anyone’s blood alcohol. But there is a wrinkle worth understanding: because Penal Code 49.01 defines the offense in breath units, grams of alcohol per 210 liters of breath, partition ratio variability is not by itself an answer to a per se breath charge in this state. Where it does real work is everywhere else in the case, on the impairment theory, on any attempt to translate the breath number into a blood alcohol level, and on retrograde extrapolation. The scientific literature on partition ratio variability has existed for decades, and it remains fair game on cross-examination of any technical supervisor who defends the number as though it measured blood.

Mouth Alcohol and the 15-Minute Observation Period

The Intoxilyzer 9000 is designed to measure alcohol from the deep lungs. It has no way to distinguish deep-lung alcohol from alcohol sitting in the mouth or upper airway. Mouth alcohol (trapped in dental work, regurgitated from the stomach, or left over from a recent drink) produces an artificially elevated reading because mouth alcohol is far more concentrated than alveolar air.

Texas DPS rules address this with a 15-minute period before the test, and the requirement is written down where anyone can read it. Texas Administrative Code Title 37, Part 1, Chapter 19, Rule 19.3(a)(1) provides that an operator shall remain in the continuous presence of the subject at least 15 minutes immediately before the test and should exercise reasonable care to ensure that the subject does not place any substances in the mouth. The same rule then says, in terms, that direct observation is not necessary to ensure the accuracy of the test result.

That last sentence is the one nobody expects. This was never a rule that required the operator to watch you. It requires presence and reasonable care, and nothing more. In practice the operator is usually typing information into the instrument and not looking at the subject at all, and that practice is consistent with the rule as written rather than a departure from it. DPS operator training treats a belch, regurgitation, vomiting, or anything placed in the mouth as restarting the 15-minute clock, but that is training and agency practice, not the text of Rule 19.3. That changes how the challenge gets framed. Arguing that the operator failed to watch continuously does not get far, because the rule does not require watching and Texas courts have followed the rule. State v. Reed, 888 S.W.2d 117 (Tex. App. San Antonio 1994), holds that presence means being near enough to be accounted present, and Serrano v. State, 464 S.W.3d 1 (Tex. App. Houston [1st Dist.] 2015), holds that an operator’s inability to recall observing does not by itself invalidate the test.

The productive attack is on the reasonable care prong and on the science underneath it. Video review in Texas DWI cases routinely shows periods that were not 15 minutes long, or that missed a belch, a burp, or regurgitation precisely because nobody was looking. What that does to a reading the instrument cannot separate from deep-lung alcohol is a forensic question, and it is why the video matters more than the offense report.

Medical Conditions That Can Skew a Reading

Several medical conditions can inflate a breath test result through mechanisms the machine cannot detect:

Gastroesophageal reflux disease (GERD) and acid reflux. Stomach contents, including unabsorbed alcohol, can be pushed back up into the esophagus and mouth during the test, producing a mouth-alcohol effect the observation period was meant to prevent.

Diabetes and diabetic ketoacidosis. Elevated ketones (particularly acetone) produce an absorption signature that older breath instruments can misread. The 9000 uses multiple infrared wavelengths to try to filter this out, but elevated ketone states can still affect the analysis in some cases.

Low-carbohydrate and ketogenic diets. Sustained ketosis from dieting raises breath acetone concentrations. In certain conditions, this can produce spurious readings or alter the instrument’s breath-sample verification.

Dental work, dentures, and bridges. Residual alcohol can be trapped in dental appliances and released during the breath sample, again producing a mouth-alcohol effect.

Respiratory conditions. Asthma, COPD, and other conditions affecting lung function can alter breathing pattern and sample composition, which influences the reading.

None of these conditions appear anywhere on the offense report. They surface only when a defense attorney asks the right questions of the client and subpoenas the medical records to support the defense.

Breath Temperature, Breathing Pattern, and Physiology

The Intoxilyzer 9000 assumes a breath sample temperature of 34 degrees Celsius. Research by Michael Hlastala and others puts the effect of deviation from that assumption at roughly six to seven percent per degree Celsius, and some studies report more. A person running a fever, a person who has just exercised, or a person whose breath is naturally warmer than average will each register higher than they should.

How the subject blows into the instrument also matters. Prolonged exhalation forces increasingly deep alveolar air into the sample, and alveolar air has a higher alcohol concentration than upper-airway air. Officers often coach subjects to blow harder and longer to get a satisfactory sample. That coaching can push the reading higher than a shorter, more representative exhalation would have produced.

Finally, where the subject is in the absorption, peak, and elimination curve at the moment of the test matters enormously. A person still in the absorption phase (whose alcohol is not yet fully distributed between blood and tissue) can produce a breath test reading that does not accurately reflect actual impairment at the time of driving. This is the basis for the rising-BAC defense, which requires retrograde extrapolation analysis. Texas has a controlling case on that analysis. In Mata v. State, 46 S.W.3d 902 (Tex. Crim. App. 2001), the Court of Criminal Appeals held that an expert offering extrapolation testimony must know facts including the length of the drinking period, the time of the last drink, and the defendant’s weight, and that a single test taken well after the driving, offered by an expert who knows almost nothing about the person, is not reliable enough to admit. Mata cuts in both directions, and it is the first case to read the moment a State’s expert starts extrapolating. Deandra Grant wrote her Master’s thesis on retrograde extrapolation, and it was later published twice in Counterpoint: The Journal of Science and the Law.

Calibration, Maintenance, and the Human Element

The Intoxilyzer 9000 is a precision instrument, and like every precision instrument it requires calibration, maintenance, and inspection. Under the DPS Breath Alcohol Program rules in Texas Administrative Code Title 37, Chapter 19, each instrument must be certified under Rule 19.2 and each Breath Test Technical Supervisor separately certified under Rule 19.5. Every inspection generates a paper trail:

Maintenance records documenting repairs, replaced parts, and service events.

Calibration records showing the reference standards used and the results obtained.

Technical supervisor credentials documenting the qualifications of the person certifying the instrument.

Subject test records showing every test run on the machine, including tests that were invalidated, retried, or produced unusual readings.

All of these records are discoverable. All of them are subject to challenge. Maintenance gaps, reference standard problems, technical supervisor qualification issues, and patterns of unusual readings on a specific instrument can all be developed on cross-examination.

It is worth understanding why a Texas breath challenge takes this shape rather than a broader one. Texas courts treat the underlying theory and technique of infrared breath testing as validated by the legislature, so a general attack on the science will not succeed. Beard v. State, 5 S.W.3d 883 (Tex. App. Eastland 1999), and Reynolds v. State, 204 S.W.3d 386 (Tex. Crim. App. 2006), hold that a certified operator need not understand the underlying science so long as the protocol was followed. What stays open is the third Kelly prong, whether the technique was properly applied on this occasion, with this instrument, by this operator, on this person. That is why the records and the video carry more weight in a Texas breath case than the physics does.

Two Different People, Two Different Jobs

The person running the instrument on the night of your arrest is a certified breath test operator, usually the arresting officer or another officer at the station. Their job is to run the sequence and to observe you beforehand. They are not chemists and they do not maintain the machine.

The technical supervisor is the person responsible for maintaining the instruments in a given area, checking their calibration, preparing the reference solutions, and training the operators. The technical supervisor is the State’s expert on breath testing, typically holding a bachelor’s degree and trained through a specialized course, and if a breath result is contested at trial this is usually the witness the State calls to establish that the machine was working. Which means the person who testifies that the instrument was reliable is almost never the person who operated it.

Nothing Was Saved for a Second Opinion

A Texas breath test does not preserve a specimen. Your breath went through the chamber and out. Nothing was captured for independent analysis later.

That is different from a blood sample where the tube still exists and can be sent to a laboratory for retesting. In a breath case the records generated that night are the only records there will ever be: the printed slip showing both sample results, the air blank and reference standard readings, the observation period documentation, and the video from the testing room.

Common Questions

You could have refused. Texas cannot physically force a breath sample out of you. Refusing carries consequences of its own, including a license suspension longer than the one that follows a failed test and a real chance the officer pursues a blood warrant instead. It is a decision with trade-offs running in both directions rather than an obvious right answer. See Can I Refuse a Breath Test?.

Two clocks start, not one. If the result came back 0.08 or higher, the criminal case proceeds and the breath result becomes the State’s central exhibit. Separately, and on a schedule of its own, an administrative license suspension begins the moment the officer serves the paperwork. That clock has nothing to do with your court date, and missing it costs you your license by default, without anyone ever hearing your side of it.

Breath and blood fail differently. A blood test measures alcohol in blood directly with no conversion step, which removes the partition ratio problem entirely, and then introduces a different set of failure points in the draw, the storage, and the laboratory analysis. Neither test is the more reliable one. See The Blood Test.

What This Means for Your Case

The accuracy of the Intoxilyzer 9000 is not a single question. It is a set of questions that have to be asked for each specific case:

Was the 15-minute period actually complied with? Video review tells the real story, not the report, and the question is whether reasonable care was exercised rather than whether anyone was watching.

Does the defendant have a medical condition that could affect the reading? GERD, diabetes, respiratory conditions, dental work (all require medical record review).

What was the defendant’s breath temperature? Fever, illness, and recent physical exertion all matter.

Was the defendant in the absorption phase at the time of the test? Timing of the last drink relative to the test is the foundation of the rising-BAC defense.

What do the instrument’s records show? Calibration, maintenance, reference standards, and subject test history are all discoverable.

What is the true measurement uncertainty of this specific reading? The single number on the ticket does not tell the whole forensic story.

Which device produced the number? A roadside screening result is not evidence of alcohol concentration. Only the station instrument counts.

How far apart were the two samples? Anything inside 0.02 passes, including a pair that straddles the legal limit.

Was there an interferent in play? Solvent exposure at work, ketosis, and alcohol-carrying medications all belong in the history.

A DWI defense that ignores any of these questions is leaving forensic evidence on the table. A defense that works through all of them can shift the scientific ground under the prosecution’s case.

The Bottom Line

The Intoxilyzer 9000 is a piece of scientific equipment, not an oracle. It produces estimates, not truths. Its output depends on assumptions that do not fit every human body, procedures that are not always followed, and records that do not always hold up under scrutiny. A breath test reading is the beginning of the forensic conversation in a Texas DWI case but not the end of it.

Breath Test Defense at Deandra Grant Law

Deandra Grant Law defends DWI and intoxication-offense cases across North and Central Texas including Dallas, Fort Worth, Plano, McKinney, Frisco, Allen, Lewisville, Denton, Rockwall, and Waco. Our team includes an ACS-CHAL Forensic Lawyer-Scientist with a Master of Science in Pharmaceutical Science and a Graduate Certificate in Forensic Toxicology, which are the credentials required to cross-examine a Breath Test Technical Supervisor on partition ratio, measurement uncertainty, and instrument performance. Deandra spends an entire chapter covering breath alcohol analysis in her book The Texas DWI Manual.

If a breath test is being used against you, call Deandra Grant Law at (214) 225-7117 or visit texasdwisite.com to schedule a confidential consultation. If you took a breath test and lost your driving privileges, you still have a 15-day ALR deadline running from the date of service of the notice of suspension. Do not wait.

Have a DWI question you want answered in this series? Submit it at texasdwisite.com! You might see it featured in a future Ask Deandra post.