The Disconnect Defense: Defeating a Breath Test Twice the Legal Limit

Posted on June 29, 2026 in Uncategorized

By Hunter Tzovarras
Chair, NCDD Trial Advocacy Task Force
The Law Office of Hunter J. Tzovarras, Bangor, Maine

With assistance from Myranda G. Sandberg
Member, NCDD
Nero DUI Defense, S.C., Eau Claire, Wisconsin

"It doesn't matter what number the machine prints out if the video shows you walk, talk, and act sober." That is what I tell clients who are sure a high breath test means certain conviction. When jurors believe a person is sober from what they see with their own eyes, they become willing to doubt a piece of paper that says otherwise.

That is the heart of what is widely known as "the disconnect defense."

A disconnect defense is built around a simple point: the number on the printout does not match the person the jury can see and hear for themselves. The machine says one thing. The video, the driving, the officer's observations, and the client's behavior say another. When that happens, the defense is not that the jury must know exactly why the machine is wrong. The defense is that the State has not proven the machine is right beyond a reasonable doubt.

I watched this happen last winter in a Maine OUI trial.

In January 2024, in Madawaska, Maine, Jesse backed up one hundred yards through the snow looking for a street he had passed in the dark. The officer smelled alcohol, saw bloodshot eyes, and asked how much he had to drink. "Two, three, four, maybe five," Jesse answered. He gave six clues on HGN, three on the WAT, and three on the OLS. The Intoxilyzer 8000 put him at 0.16.

Despite the field sobriety clues and a result at twice the legal limit, I knew the case could be won, and that it had to be tried. It was winnable because over a full hour of body-camera video, Jesse does not look like a man at twice the legal limit.

This was a pure disconnect defense. We had no operational or procedural attack on the breath test, and no medical explanation to discredit the result. The theory was simple: the test must be wrong because the jury can see for themselves that Jesse is not impaired.

Start With What the Jury Can See

From opening through closing, the strategy never changed. The job was to show the jury that the number on the printout does not match anything else they can see and hear about Jesse.

It began with the first words of the opening statement:

Jesse is not guilty of OUI, because four or five beers over several hours does not get anyone close to twice the legal limit. If this machine were accurate, Jesse would have tested at 0.06 or below. Everything you see with your own eyes and hear with your own ears in this trial will show you that Jesse was sober. The number this machine printed is simply wrong. That is why Jesse is not guilty.

That is the structure of the defense. Before attacking the machine, give the jury a person to compare it to. The more signs of sobriety the jury holds, the more willing they are to question the field sobriety clues and eventually the number itself.

So the first job is to collect every sign of sobriety on the video and leave none on the table.

• Did the client drive safely?
• Did he pull over safely?
• Did he answer questions appropriately?
• Was his speech clear?
• Did he follow directions?
• Did he have good motor skills?
• Did he walk and stand well under the circumstances shown on video?

Those facts matter because they let the jury say what the defense wants them to say: this does not look like twice the legal limit.

Put the Field Sobriety Tests in Context

The next step is dealing with the field sobriety tests. Jurors often hear "clues" and assume those clues prove impairment or prove a BAC over 0.08. But the officer's own training can show something different.

In Jesse's case, I started where the officer's own training starts. The 2018 NHTSA manual concedes the tests flag drivers below the legal limit: "It strongly suggests the SFSTs also identify BACs at 0.04 and above." Once the jury understood that clues can appear below 0.08, the field sobriety tests no longer had to mean what the State wanted them to mean.

Then I gave the jury a number for Jesse below the legal limit, built entirely from the officer's own training. NHTSA assigns roughly 0.02 BAC to a standard drink for an average man and an elimination rate of 0.015 per hour. Run against Jesse's drink count and timeline, the math lands under the limit, and it doesn't take an expert to get there.

That number matters. In a high-test case, the jury needs more than "maybe the machine is wrong." They need a competing explanation – an actual, positive theory of innocence – they can understand and repeat in deliberations. For Jesse, the number was 0.06 or below.

HGN also matters, but, like any SFST, only if the officer did it correctly. In Jesse's case, HGN fell out because the officer administered it improperly (a good topic for another blog post). On the two physical tests, Jesse looked good on video even though he was walking and standing in the snow. That became part of the same disconnect: the clues did not match the person the jury watched.

Make the Machine the Outlier

Only after building the signs of sobriety did I move to the machine itself.

I put a photograph of the Intoxilyzer 8000 in front of the jury and walked the officer through what he could not defend: that the manufacturer warrants nothing, that the source code is secret, and that he could not say how the machine arrives at the number it prints.

The questions were simple:

"The machine doesn't weigh any alcohol, does it?"

"It isn't specific to alcohol, is it?"

"It really measures how much light passes through Jesse's breath, right?"

"And Jesse could have other substances in his breath, ones that are not alcohol, that absorb light too, couldn't he?"

Every answer helped. Agreement made the point; "I don't know" made it better. Either way, I could argue in closing that the machine is a black box that does not match the man the jury watched for an hour.

It also helped that Maine was then retiring the Intoxilyzer 8000 in favor of the 9000. CMI built a "new and improved" model, I reminded the jury, and you improve something for a reason.

The goal was not to prove exactly why the machine was wrong. The goal was to show the jury that the State's number was the one thing in the case that did not fit.

Bring the Disconnect Home in Closing

Closing is where the signs of sobriety come together into reasonable doubt. We ought to leave the jury with a few plain images and taglines they can carry back into the room.

The first is borrowed, shamelessly, from NCDD Regent & Faculty Adam Nero, who borrowed it, shamelessly, from NCDD Member John Menzel, and who he borrowed it from is unclear and you'll have to ask him. The image is three red balloons and one blue one, under the line "One of These Is Not Like the Others," yes, as in the song from Sesame Street. The red balloons are Jesse's driving, his demeanor, and his physical performance. The breath test is the blue one. It sticks out like a sore thumb because it is the only piece of evidence purporting to point to alcohol impairment.

That is the whole case in one image.

Without a breath-test expert to testify, we can still lean on the jurors' common sense. A person at twice the legal limit cannot do what Jesse did on that video. We can walk the jury through all of it: the safe driving, the safe stop, the clear speech, the good motor skills, the clear mind, the ability to follow directions, and the performance on field sobriety tests under tough conditions.

In Jesse's case, I also made the point that machines are not error-free. Everyone knows that, from the phone in their pocket, to the bathroom scale, to a multimillion-dollar airplane. A breath-test machine does not become infallible just because the government uses it. Then I gave the jury the Defense's theory-of-innocence number again. We should always propose our own number to the jury. For Jesse, it was 0.06 or less.

Finally, I reminded the jury of the court's instruction that they alone decide whether to accept or reject the reliability of the breath test, and that they are free to reject it. Against the State's argument that the machine was approved and working as designed, I left them with this:

We can know something is wrong without knowing why. Jesse does not have to prove why the machine is wrong. The burden is on the government to prove, beyond a reasonable doubt, that it is right. If there is any reasonable possibility that it is wrong, even a small one, you must find Jesse not guilty.

This is a human decision. Whether a fellow citizen is guilty of a crime is not a judgment we hand to a machine none of us can explain. We can, and arguably should, end every closing the same way, by directly asking the jury for the NOT GUILTY verdict or verdicts we want from them. In Jesse's case, that is the verdict they brought back.

A high BAC reading does not have to mean a guilty verdict. When we help the jury see for themselves that the number does not fit the person, through the lens of a positive theory of innocence and sobriety, the number is the thing that breaks.