Are the breath-alcohol testing machines used by law enforcement officers reliable? The answer, admittedly, is: sometimes. Scary isn't it?
First of all, most law enforcement officers will admit that the machine, if working properly, has a margin of error that they deem "acceptable" of .02 grams of alcohol/210 Liters of breath. In turn, if the machine is working exactly right, the resultant test score of .08 grams of alcohol/210 Liters of breath may actually be a .06 or a .10.
Then, there are testing issues. The laboratory or non-laboratory-like cleanliness of the testing facility can certainly impact the score. Some experts estimate that the cleanliness of the testing facility can affect the breath-alcohol test score by as much as .06!
In addition, the machines test themselves for accuracy, which is to say that there is no external apparatus that pretests the machine for accuracy. Instead, the machine itself tells the test administrator if it is working properly, and there is no additional test that insures that the machine's fail indicator is not failing.
In addition, the machines are supposed to be certified by a key operator and re-certified each time they are moved. The machines themselves are about the size of a 1989 IBM typewriter. Janitorial personnel routinely move them to dust under them, yet they are only certified by a key operator (in most instances) twice a year.
There are many, many other problems that affect the reliability of the breath-alcohol testing machine. In brief, two independent studies of the machine's source code have revealed more than a thousand errors in the source code. In turn, even if the machine is working right, it will provide an inaccurate score periodically due to machine error.
The problem is that in order to point out the major issues with the breath-alcohol testing machine, an expert witness is required by the Court. Expert witnesses are very, very expensive.