concurs and dissents.
¶42 I concur with the Court’s conclusion on issues Two and Three. As to Issue One, I strongly disagree with our conclusion that PBT results should be admitted as substantive evidence of guilt in DUI proceedings.
¶43 As acknowledged by the Court, this is not a new issue for us. In 1997, we determined in Strizich that the results of a preliminary breath test (PBT) (a/k/a preliminary alcohol screening test, or PAST) are just that-preliminary. In that case, we reviewed substantial testimony offered before the 1995 Legislative Session on the technological reliability of the testing instruments used by officers in the field. We concluded that ‘the results of a preliminary breath test ... are not substantive evidence of the amount of alcohol present in a person’s body, but instead are an ‘estimate’ of alcohol concentration for the purpose of establishing probable cause to believe that a person is under the influence of alcohol prior to making an arrest....” Strizich, 286 Mont. at 12, 952 P.2d at 1372. This conclusion was based on reputable testimony that the instrument when used in the field is subject to several unpredictable variables based on physiological, *288instrumental, and operator-induced factors. Strizich, 286 Mont. at 12, 952 P.2d at 1372.
¶44 This issue was again before us in 2003. See Weldele and Crawford. In both cases, we once again determined that ‘preliminary breath testing in the field as it is currently administered remains statistically unreliable.” Weldele, ¶ 57. See also Crawford, ¶ 18. As recently as November 2004, this Court determined that a district court abused its discretion by admitting the results of a defendant’s PBT. Snell, ¶ 38. In Snell, we explained that the State had ‘presented virtually the same evidence it did in Weldele, with the exception of one additional study of the Aleo-Sensor III from the Idaho Department of Health and Welfare which determined, among other things, that weekly calibration of the instruments helps to ensure accuracy....” Snell, ¶ 37. We noted in Snell that Montana calibrates its machines on a monthly basis. Snell, ¶ 37. We held in Snell that
After a review of the record, we conclude the State failed to refer to new evidence that demonstrates the reliability of the AlcoSensor III and we are not satisfied the evidence presented sufficiently established PAST accuracy and reliability so the State could use the PAST for more than probable cause purposes.
Snell, ¶ 37.
¶45 While we invited the State to present evidence at any time that establishes the scientific reliability and accuracy of the PBT results {Weldele, ¶ 57), in my judgment it has not done so. In fact, I would conclude that the evidence introduced in this case reaffirms the conclusions we reached in our previous decisions rejecting PBT results due to their unreliability as indicators of guilt.
¶46 In the case sub judice, two expert witnesses testified at length about the history, operating procedures, and, more generally, the strengths and weaknesses of the Alco-Sensor III: DeWayne Beckner (Beckner), a twenty-five year veteran of the Los Angeles County Sheriffs Crime Lab who specializes in forensic alcohol tests, testified for Damon, and MacQuorn Forrester (Forrester), the Chairman of the Board of Intoximeter, the company that developed and manufactures the Alco-Sensor III, as well as other alcohol detection instruments, testified for the State.
¶47 Beckner initially noted the distinction between an instrument used for ‘probable cause purposes” and one used for “evidentiary purposes.” He explained that when using an instrument for probable cause purposes, the operator is merely getting a preliminary assessment of the subject’s condition to aid him or her in determining *289whether the subject might be intoxicated. On the other hand, with reference to any instrument to be used for evidentiary purposes, he said: ‘You are determining to a very precise concentration the alcohol content of your subject’s breath or blood; and you’re recording this data; you’re complying with whatever regulations that might exist in your jurisdiction; you’re using the highest scientific protocols available; and you are doing it all under the umbrella of properly administered scientific tests.”
¶48 Beckner opined that the Alco-Sensor III, which “was designed for preliminary alcohol screening purposes,” was certainly acceptable as a “probable cause screener,” but that to use it as an “evidentiary instrument” would require applying the same criteria to the collection of data and training of operators as applied to evidential instruments (e.g., Intoxilyzer 5000). He pointed out that evidentiary station-based instruments, such as the Intoxilyzer 5000, generally cost between $5,000.00 and $6,000.00 while the Alco-Sensor III costs approximately $500.00. Beckner stated that the significantly more expensive instruments provided a much greater level of reliability because the machines contain substantially more sophisticated technology.
¶49 Beckner provided examples of some of the controls available in the station-based units that are not available in the hand-held units. He noted that white bread in the mouth of a subject could render a positive reading in an Alco-Sensor III unit because of the yeasts and compounds contained in the bread. He explained, however, that infrared station-based instruments have the capacity to look for organic compounds that could interfere with Alco-Sensor III readings, such as acetones, ketones, acyl aldehydes, and many others. He also testified that the more sophisticated stationary machines detect the presence of “mouth alcohol” and render an invalid test until such mouth alcohol dissipates. He stated that while officers are instructed to conduct Alco-Sensor field tests only after a minimum fifteen-minute deprivation time to avoid an inaccurately high reading based on the presence of mouth alcohol, some officers fail to do so. He also testified that for some people with dental appliances or other dental conditions, fifteen minutes is an inadequate time to eliminate the risk of measuring mouth alcohol. As a result, these persons when tested exclusively on the Alco-Sensor III without the safeguard of mouth alcohol detection, may get an incorrect high test result.
¶50 Beckner also attributed greater reliability to the station-based units because they are not portable and, therefore, are not subject to the stresses of portability. He explained that many of the Alco-Sensor *290units used by California officers are subject to very rough handling-they are left on top of patrol cars, they are dropped, damaged, ‘bounced around” in a squad car and stored inappropriately. He maintained that such treatment affects the reliability and calibration of these instruments.
¶51 Further according to Beckner, once an instrument is out of calibration, it should not be used. He explained that most modern instruments have software that prohibits the instrument’s use when it is out of calibration but ‘if the calibrations are done manually, and it’s up to a human being to check the value of the calibration versus the value of the standard, then it can be a problem.” He testified that the Aleo-Sensor III must be calibrated manually. He also offered that as the fuel cells within the Aleo-Sensor III instruments age, they become less sensitive and less reliable.
¶52 Beckner testified that both environmental conditions and operator practices in the field can influence the results of an Aleo-Sensor III test. He reported that while ambient temperature generally is not a problem, residual condensation could occur when an instrument is used in “very, very cold” ambient air. He stated this could cause a “dramatically’ inaccurate reading. As an example, he explained that a person could ‘be an .08 when [they were in reality] a .00.” Under these circumstances, to guarantee an accurate reading, the operator would have to warm the unit or run “air blanks” through it.
¶53 As for operator-controlled factors, he pointed out that an officer could do a “manual” capture of a subject’s breath rather than waiting for the instrument to complete the “automatic capture” and this could cause an inaccurate test result. This could occur intentionally as a result of impatience or inadvertently due to the stress of the situation in which an officer finds himself or herself. Another source of operator error is failure to utilize an instrument’s capability to record and store information and later download it to validate the date, time and test result. Beckner stated that many officers “merely write down what they claim to have seen, and then the evidence disappears, as opposed to collecting the evidence, memorializing it, and validating that it actually occurred.” Beckner believed that most officers honored the fifteen-minute deprivation time and the absolute rule of using a new, dry, clean mouth piece for each test, but recognized that failure to adhere to this strict protocol would affect the test results.
¶54 Beckner stated that, if he was going to use PBT results for evidentiary purposes, an element of his strict protocol would be to take at least two samples. He claimed he would never use a single PBT *291result for evidentiary purposes. He explained that “evidence sampling variation” is frequently a factor, meaning that a second PBT sample taken two minutes after the first sample will commonly reveal a .02 and .03 difference. He asserted that tests in a crime lab to be used for criminal prosecution would never go out without a confirmatory test attached to it and that frequently such tests are done “two or three or four” times to make sure they are right.
¶55 Beckner noted that the Alco-Sensor III in a pristine laboratory is very accurate, but once it is taken outside to the field, “that’s where the uncertainty start[s] to creep in to how accurate the reading may or may not be.”Beckner further testified that Montana’s protocols on the use of the Alco-Sensor III “need to be tightened up,” and that Montana is not using “really high level protocols.” For example, Montana does not require the taking of more than one sample.
¶56 MacQuom Forrester, who bought Intoximeter from his father more than forty years ago, testified for the State. Much of his time during these decades was spent in the research and development department. He testified at length about the history of the Alco-Sensor III technology and proper operating procedures. Forrester stated that training and good observance of procedures is “essential” to the accurate operation of the Alco-Sensor III. He opined that 99% of inaccurate readings were the result of operator error, or what he called ‘the human element.” He recognized that there are times when a field sample will deviate beyond the acceptable 10% margin of error. Examples he provided to explain such a deviation were the ‘incompetence level of a police officer ... [or] the vindictiveness of a police officer that’s been spit on.”Forrester also pointed out a number of operator errors that could result in inaccurate readings, including, but not limited to, using the machine when its fuel cell is too cold, using the machine too many times in a short period of time, communicating through a police radio transmitter during a test, and not allowing an appropriate deprivation time before drawing a breath sample.
¶57 Notably, Forrester candidly admitted that there have been no recent technological advances to the instrument that would render it more accurate or reliable under field conditions. To the contrary, Forrester testified when asked if the company has made any technological changes since 1980 that would render the instrument more accurate, ‘Not the accuracy, but ... it may be a little faster in response. ... But not in terms of the analytic process, the outcomes to come up with the final reading, no. The sampling system is identical.”
*292¶58 Forrester acknowledged that while there have been substantial changes to the Aleo-Sensor III operation manual, these changes were the result of criticism from law enforcement agencies, as opposed to being the result of any advances in science. An example of such a change was that in an earlier manual the company recommended, “Where important evidence is being collected,... a standard [should] be run prior to a test to establish the accuracy of the unit for the day.” According to Forrester, this recommendation was removed from the manuals because states “regulate the process.” In essence, Forrester admitted that operating instructions will vary from state to state, and that these altered protocols have not been independently evaluated or subjected to peer review. It is therefore impossible to say with any degree of objective certainty whether the procedures governing use of the instrument, as revised in Montana or elsewhere, have enhanced or diminished the instrument’s reliability.
¶59 No doubt this complete lack of inherency is part of the reason that the vast majority of states still exclude the results of the PBT or PAST as evidence of guilt. While the District Court found, based on Forrester’s unsupported testimony, that “the Aleo-Sensor III is used exclusively in at least 17 other states,” an actual study of the 50 states’ use of this instrument (per Amicus Curiae Montana Association of Criminal Defense Lawyers ‘Summary of PBT Rules, By Grounds”) reveals that only five states allow Aleo-Sensor III test results to be admissible as substantive evidence of guilt. And in general, the states that allow such test results for evidentiary purposes impose more stringent protocols than does Montana1.
¶60 As we noted in Weldele and Crawford, jurors are inclined to give more weight to evidence that is presented under the imprimatur of science than to non-scientific evidence. Weldele, ¶ 55, Crawford, ¶ 18. See also State v. Weaver, 2005 MT 158, ¶ 43, 327 Mont. 441, ¶ 43, 114 P.3d 1039, ¶ 43 (Cotter, J., dissenting). A numerical value from a machine, served up with a decimal point, is a known quantity that jurors can grab onto; it easily beats the amorphous concept of reasonable doubt. This tendency to accept scientific data is especially *293significant in a DUI case because the statistical result is literally the difference between a determination of guilt or innocence. While in other types of cases, scientific results may have relative or comparative value, in DUI cases, the result has absolute repercussions. The law establishes a definitive bright line between criminal guilt and innocence at a .08 intoxication standard. As we stated in Weldele, ‘tb]ecause the statistical result of such a test could upon admission at trial determine guilt, the test result must be demonstrably accurate.” Weldele, ¶ 57. While it is frequently necessary to accept a ‘inargin of error” percentage when utilizing quantifying instruments, the margin of error currently attributed to the Aleo-Sensor III in combination with the possible inaccuracies attendant to its portability and use in field conditions simply render the machine too unreliable for use in establishing criminal guilt, especially when the more accurate and reliable Intoxilyzer 5000 is readily available.
¶61 I am very concerned that acceptance by this Court of PBT results will lead to a reduced use, if not the extinction, of the more reliable and accurate infrared machines such as the Intoxilyzer 5000. Obviously, the Aleo-Sensor III is small, portable and cheap. It is easy to use, the protocols are more relaxed than are the Intoxilyzer 5000 protocols, and the results are readily available. Under the Court’s decision, Intoxilyzer 5000 test results will be at best superfluous, and at worst, an outright inconvenience when at odds with the PBT. It will simply be more expedient to do away with the Intoxilyzer test altogether. And with the demise of the Intoxilyzer test will come the demise of the guarantees of accuracy we used to require in cases such as this.
¶62 As a result of our decision the results of the Aleo-Sensor III PBT will now be admitted as a matter of law in future DUI cases. There will be no argument about the obvious inherent unreliability of the instrument; there will be no testimony from experts like DeWayne Beckner; there will be no Daubert2 hearings-we have settled the issue of admissibility once and for all.
¶63 At best, and assuming it even occurs to the accused’s attorney to do so, defense counsel will be able to subpoena the administering officer and the authorities who maintain and calibrate the PBT instrument so as to challenge its use and operation in the particular field setting of the case. Experience teaches, however, that this sort of attack will not be effective. Trial courts will admit the PBT evidence *294because we have said it is admissible, its scientific validity having now been decreed as a matter of law. That is all that will matter.
¶64 Given the litany of factors identified by experts for both sides that underscore the unreliability of the results obtained from the AlcoSensor III PBT, I am at a loss to understand why we feel so constrained to buck the well-reasoned national trend and admit these results as substantive evidence of guilt. Nothing has changed since Strizich, Weldele, Snell or Crawford to deem these machines magically more trustworthy than they were just a few short months ago, or to justify this about-face in our jurisprudence. I therefore dissent.
JUSTICE NELSON joins in the concurrence and dissent of JUSTICE COTTER.Arizona requires, among other things, that a confirmatory duplicate test be performed. A.R.S. § 23-1323; California requires an accuracy test of the instrument every ten days or every 150 tests, whichever is sooner. 17 Cal.Admin. Code §1221.4; Idaho requires that the test be conducted in accordance with specified methods and that an accuracy test is performed every seven days. State v. Smith (Id. App. 1997), 947 P.2d 1007 and Snell, ¶ 37; Ohio requires weekly calibration, Ohio Admin. Code §3701-53-04.
See, State v. Clifford, 2005 MT 219, 328 Mont. 300, 121 P.3d 489, (Nelson, J., concurring).