Breath-Alcohol Analysis: How Reliable Is It?

By Gerald D. Simpson, Ph.D.

In a 1983 law review article, Stephen G. Thompson observed the following:

“Modern criminal justice is premised upon the requirement that a criminal defendant be proved guilty beyond a reasonable doubt before punishment can be meted out. This standard of proof is severe; its severity is based upon a collective societal judgment that the risk of error be borne by the state. As fundamental and unquestionable as this principle may seem, it is frequently tested when the interests of society appear urgent, immediate, and identifiable. In these instances, society often creates policies and systems which threaten the presumption of innocence.”

Breath testing is a good example of the use of scientific evidence that routinely deprives suspects and defendants of the presumption of innocence and results in wrongful convictions as well as unwarranted guilty pleas.

The reason for this is that breath testing as now employed does not accurately reflect the true or actual value of alcohol concentration in the venous blood or even in the breath of a human subject.


Current scientific research, published in respected peer-reviewed journals, now shows that results from breath alcohol analysis or breath testing are not sufficiently reliable for use in our courts. This represents a huge problem because breath testing has been used widely for this purpose for over 50 years and continues in widespread use on a daily basis.

What has now been demonstrated is that three important aspects of breath testing are faulty or even downright wrong:

  1. The very basis for breath testing in humans is the theory that ethyl alcohol (ethanol) in the blood is in equilibrium with the alveolar air in the lungs and that by measuring the concentration of ethanol in the end-expired air sampled by a breath machine, the amount of ethanol in the blood or breath can be reliably determined. Recent work by Michael Hlastala at the University of Washington Medical School in Seattle has shown that there is no such equilibrium and that breath testing therefore cannot work.
  2. Evidential breath testing has a very large margin of error, a minimum of +46%, meaning that for a given individual the breath test result can be anywhere from 0 to 46% higher than the actual amount of alcohol in the blood. This is the minimum amount of error that is indicated by the most recent scientific research. The actual amount could well be far greater.
  3. It has been recently shown that the “calibration” method used to ensure reliable performance of evidential breath machines does not work. The method used to “calibrate” these machines on a regular basis deals with only a fraction of the total error or uncertainty involved in a breath test result for a given individual; it simply ignores the major sources of error involved in results from evidential breath machines, e.g. error that occurs from assuming that the blood/breath ratio is 2100:1. Consequently, the actual amount of error in a breath test result for a given individual is unknown.


Despite the fundamental scientific shortcomings of breath testing as used in criminal and civil cases, our courts continue to consider such evidence as an acceptable indicator of the inability to drive safely.

One issue that should be taken to the U.S. Supreme Court involves the destruction of evidence when breath testing is are given.

When a suspect is given a breath test, the breath sample is discarded, even though the technology to save the breath sample has been readily available for many years. This issue was taken to the U.S. Supreme Court in a California case called Trombetta many years ago (1984) and the court ruled that breath test samples need not be saved.

It is time to revisit this issue because research done in the 1990s makes it clear that breath testing is far less reliable than blood testing, and far less reliable than the Trombetta court was led to believe, yet it is blood samples that are retained by forensic laboratories and breath samples are discarded.

There are at least 100 volatile compounds (other than ethyl alcohol) that can cause false readings on a breath test device.

This is a compelling reason why the breath sample should be saved for reanalysis. A simple gas chromatography analysis can determine if any other compounds are contributing to, or are responsible for the entire reading. The prosecution must provide an unequivocal identification of ethyl alcohol and show that it is responsible for the entire reading produced by the breath test device. Without this, the evidence would appear to be insufficient for a reasonable juror to conclude guilt beyond a reasonable doubt.

Because a criminal charge is involved, regulation should be at least as strict as it is for medical or clinical laboratories for which 95% confidence limits are used to meet a standard of reasonable medical certainty.

Breath testing is actually a clinical chemistry analysis because alcohol concentration is being measured in a human body fluid (breath). Consequently, when used for legal purposes in criminal cases, regulations should require at least 95% confidence limits (some have argued that 99% confidence limits are appropriate to meet a standard of beyond a reasonable doubt).

Consider a political poll in which one candidate gets 49% of the vote and the other 51% of the vote, and the margin of error in the poll is +/- 4%. Since the result is within the margin of error, the race is too close to call. Confidence limits are part of how the margin of error is determined.

If one wants to be 95% certain that a subject’s blood alcohol concentration or breath alcohol concentration are not being over-stated by the breath machine result, an appropriate correction factor must be used.

For example, most statutes now criminalize driving with a BAC of 0.08%, and to be 95% certain that the 0.08% limit is in fact exceeded, a person would have to have a test result 40% greater than 0.08%, which is 0.112%. For 99% confidence limits the result must be 46% greater than 0.08%, or 0.117%.

At best, however, statutes and courts allow only +/- 0.01% for the margin of error in breath test results.


Drivers need to know what their rights are in order to minimize the chances of wrongful arrest and conviction.

A driver has the right to politely refuse to answer any questions about drinking.

It is likely that the officer will persist in an attempt to get the driver to admit to alcohol consumption to make their case stronger. There is, however, little chance that a driver’s explanations will keep an arrest from happening, so it is better to remain silent about anything having to do with alcohol consumption.

A few things to keep in mind about field sobriety tests:

  • If there are factors in play that compromise field sobriety test performance, such as poor balance or nystagmus due to injuries or prescription drugs the officer must be told about them in order to include them in his report.
  • Hyperventilation prior to the breath test will significantly lower the test result.
  • If the driver has had exposure to paints or solvents or is diabetic or has asthma, the officer must be asked to include these in his report as well.
  • Refusal of a blood or breath test can be helpful in some jurisdictions and harmful in others for avoiding conviction at trial, but usually results in suspension of the suspect’s driver’s license.


The current status of breath testing is, as it has been for some 50 years, about the same as the status of DNA Fingerprinting was when if first began to be used in our courts. Forensic scientists adopted this methodology and began presenting it in courts as powerful evidence to identify suspects who left their DNA at the crime scene. At this time, like breath testing now, there was virtually no oversight or outside regulation of DNA Laboratories. Like breath testing, outlandish claims of accuracy and probability of a match were commonplace by those in the forensic science community, even under oath.

As time passed, very capable and informed defense attorneys were able to raise enough doubt about the claims that were being made that evaluation of the method by the National Academy of Sciences (NAS) was initiated. As a result, most of the scientifically unacceptable aspects of the procedures and methods used by DNA Laboratories were corrected, based on the recommendations of NAS committee members.

This is precisely what must be done with breath testing.

NAS consists of members of the main-stream scientific community, not forensic scientists. Indeed, forensic science is not one of the disciplines even represented in the National Academy of Sciences. NAS must be enlisted to evaluate the scientific merit of breath testing methods and procedures. If this is not done, the 50 year history indicates that neither the forensic science community nor our courts will solve the problem.

An immediate fix for the problems with breath testing is to inform jurors and judges that there are large unknown uncertainties in breath test results, and this must be factored into decisions of innocence or guilt of drunk driving. Long term solutions await development of direct, non-invasive blood alcohol tests.

Breath testing is now and will always be too unreliable to use for legal purposes, and when evaluated by NAS, this will eventually be shown to be true.

This information is excerpted from the forth-coming book: CSI/Forensic Fraud Cover-up by John Kelly who can be reached at:

Not an NMA Member yet?

Join today and get these great benefits!

Comments are closed.