Breath analysis is probably the most commonly used technique by law enforcement to attempt establishing that a suspect was driving with a prohibited amount of alcohol in their system (i.e., .08 or above -- or simply impaired "to an appreciable degree") -- the reliability of this test has been called into question.

Some scientists and experts contend that breath tests, as they are currently administered by police, are very inaccurate [as far as their ability to accurately measure blood alcohol content (BAC)]. Some courts have even thrown out breath test results because of their unreliability.

In 2002, the Washington Supreme Court ruled that a woman could not have her driver's license suspended because of reliability problems associated with the alcohol breath test.

The woman challenged the administrative suspension of her driver's license, by taking issue with the reliability of the thermometers used in breath-test machines.

Thermometers are used to test the machines' accuracy. Operators test the machines by taking a reading from an alcohol/water mixture. The mixture must be kept at 34 degrees centigrade, which is where the thermometer comes in. The Supreme Court ruled that because evidence was not produced to establish that the thermometer used on the machine was certified as reliable, her license could not be suspended.

At least one court has even reversed DUI convictions on the grounds that breath tests are inherently unreliable. In State v. McGinley, 550 A.2d 1305 (N.J. Super. 1988), the New Jersey Superior Court, Law Division, considered the consolidated appeals of four defendants whose convictions involved Breathalyzer tests.

Although noting that the New Jersey Superior Court had essentially taken judicial notice that the Breathalyzer models "900 and 900A are scientifically reliable," the court nevertheless felt free to consider new scientific evidence not previously available.

The Court said that new scientific evidence showed:

     1.     The breathalyzer is designed to test persons having a 2100/1 blood-breath ratio. Such ratios in fact vary from 1100/1 to 3200/1.  The variance can produce errors in test results. In fact, high readings were taken in 14% of the population.
     2.     The temperature of the machine itself varies, thus affecting test results.
     3.     Body temperatures in fact vary between human beings, also affecting test results.
     4.     Hematocrit, or the solid particles in whole blood, vary (particularly between males and females, which also affects test results).

Even with the inherent unreliability of various breath testing machines, errors can be magnified when police fail properly follow procedures, such as in calibrating a machine, obstructing the port, testing blank specimens, or making sure a suspect is not affected by his or her ambient environment.