A Phoenix Police Officer stopped Robert Gosnell for speeding and then arrested him for driving under the influence of alcohol in violation of A.R.S. § 28-692(A) and for driving with a blood alcohol content of .10 percent or above in violation of A.R.S. § 28-692(B).
At the police station, the arresting officer administered a breath test to Gosnell using the GCI Mark IV machine. This machine analyzes the breath blown through it and indicates the result by digital display and by graph. In Gosnell’s case, the digital display registered .17 percent blood alcohol content, while the graph registered between .18 and .19 percent.
When the Mark IV processes the breath sample, it changes it in such a way that the suspect cannot check the result by conducting his own test on the sample. Consequently, in Baca v. Smith, 124 Ariz. 353, 604 P.2d 617 (1980), the Arizona Supreme Court held that officers must, upon request, preserve a second sample. Suspects then have the opportunity to use the test results from the second sample to contest the results of the first test.
Gosnell requested a second sample. Matich performed the test on a field collection *112unit rather than on a Mark IV. The field collection unit consists of a tube through which the suspect blows. A valve at the end of the tube makes a noise as air passes through it. This helps to insure that a deep breath is exhaled and collected in the device, a factor important to the validity of the sample. When the officer believes that he has obtained a valid sample, he crimps the tube into three compartments. The analysis of the tube’s contents takes place at a later time.
A private, state-certified laboratory analyzed Gosnell’s sample. The test produced a result of .10 percent blood alcohol content. Gosnell submitted a motion in limine to exclude both tests. He argued that the state could not establish the foundational requirements for the admission of the tests because of the discrepancy between them.
At the hearing in city court, the arresting officer testified that he had performed both tests in accordance with established procedures. He stated that he had no reason to believe that either machine was functioning improperly. He noted, however, that operating the field collection unit required him to concentrate on several factors at once. He stated that this made it difficult to insure that the suspect was blowing through the tube correctly when he crimped it.
Lucian Haag, the state-certified criminalist who analyzed Gosnell’s sample, testified that his tests indicated that the tube remained properly sealed until the time that he tested its contents. A microscopic examination of the crimp revealed no leaks. In addition, both of the compartments tested produced the same reading. Haag testified that this would not have occurred had there been a leak. He also stated that the blood-alcohol reading of a properly-sealed tube would not change as the result of external conditions or the passage of a short period of time.
Edward Raphael, a City of Phoenix criminalist, testified that the Mark IV machine in question was tested nine days before and five days after Gosnell’s test. The machine was working properly on both occasions. With regard to the discrepancy between the Mark IV’s digital display and its graph, Raphael testified that the Arizona State Department of Health considered a test valid if the two results were within one unit of each other. He believed that the discrepancy in this case occurred because the graph paper was not aligned so that the base line was at zero. It was possible to adjust for the misalignment, he stated, by subtracting the base line’s height above zero from the figure the graph indicated. When this was done, the discrepancy between the display and the graph was within the acceptable range.
Both experts testified that the discrepancy between the Mark IV and the field collection unit tests was so great that at least one of the test results had to be wrong. However, neither expert could determine which test was incorrect because there were variables that could invalidate the result of either one.
The presence of mouth alcohol as a result of burping could produce a higher Mark IV reading. If the suspect took only a shallow breath, paused while blowing, or inhaled, the field collection unit test could have a lower result. The established procedure of observing the suspect for some time before administering the test was aimed at preventing the problem of mouth alcohol. The procedure of listening closely to the sound of the air passing through the valve was intended to prevent problems with the field test. Officer Matich testified that he followed the established procedures. However, he did note that he had trouble listening to the passage of air through the valve while he was crimping the tube.
The judge granted the motion in limine to suppress the results of both tests. He based his decision upon Rule 403, Arizona Rules of Evidence, and stated as follows:
The confusion engendered by all of the evidence with respect to the validity of the respective tests leads me, at least, and I think any fact finder and any jury to nothing but a state of speculative confusion.
The state appealed the ruling to the superior court by way of a special action. The superior court affirmed the ruling and this appeal followed.
*113On appeal, we must determine whether the judge’s decision to suppress the test results constituted an abuse of discretion. The defendant argues that the inability of the experts to scientifically determine which result, if any, is correct creates a danger of prejudicing and confusing the jury. He asserts that this danger makes the test results inadmissible under Rule 403, Arizona Rules of Evidence.
First, we note that the statutory foundation for the admission of the test results prescribed by A.R.S. § 28-692.03 was met. This militates in favor of the state even if it is not conclusive. It also occurs to us that Rule 403, by its language, applies only to cases in which there may be “confusion of the issues.” Rule 403, Arizona Rules of Evidence. There is essentially only one issue in this case and the admission of the conflicting test results does not create the risk of injecting extraneous issues into the proceeding. Compare United States v. Marvin, 720 F.2d 12 (8th Cir.1983) (holding that evidence of the defendant’s sexual behavior was not at issue in the case and would have diverted the jury’s attention from material issues) with United States v. Torniero, 735 F.2d 725 (2d Cir.1984) (holding that a judge could not exclude evidence relating to insanity defense solely because the jury would be confused by the clashing opinions of experts).
We need not, however, determine how narrowly we should read Rule 403 because there is another reason why the motion in limine should not have been granted. We reject the contention that there is no way to reconcile the discrepancy between the two tests. There was evidence that the presence of mouth alcohol could artifically elevate a reading and the arresting officer testified that he could not be certain that Gosnell was exhaling properly when he collected the second sample. These are the kinds of problems juries are called upon to resolve all the time. Indeed, the very purpose of collecting a second breath sample is to allow the accused to cast doubt upon the validity of the test results. We do not believe that every time the samples vary neither can be used. It was an abuse of discretion to grant the motion in limine.
Remanded with directions to deny, the motion in limine.
SHELLEY, P.J., and HAIRE, J., concur.