This case concerns the validity of the approval by the state toxicologist of a device to measure breath alcohol content. That measuring device is known as a BAC Verifier DataMaster machine (DataMaster). This defendant, charged with driving while intoxicated, along with similarly charged defendants in other pending cases, moved to suppress the test results, each of which indicated a breath alcohol content violative of RCW 46.61.502. A number of cases were consolidated; a hearing on the suppression challenge to DataMaster results was held in the Bellevue District Court of Judge Joel A.C. Rindal.
Judge Rindal denied the motion to suppress. Defendant appealed to superior court pursuant to the Rules for Appeal of Decisions of Courts of Limited Jurisdiction. The Superior Court reversed. We granted discretionary review and reverse the Superior Court.
While this case has great significance in the use of the breath alcohol testing devices in use by law enforcement throughout the state, the legal principles upon which the decision rests are well established.
*829The legal principles involved are summarized as follows: First, the Legislature has validly delegated to the state toxicologist the authority to approve methods of analysis of a person's blood or breath to determine alcohol content. Second, the courts have inherent power to review the toxicologist's approval to assure that his exercise of his delegated authority is not exercised in an arbitrary and capricious manner. Third, the review on appeal, here and in the superior court, is governed by the RALJ standards contained in RALJ 9.1.
We review these principles in more detail, starting with the delegation of authority.
The Legislature has defined the crime of driving under the influence of intoxicating liquor as driving while a person has a specifically stated breath or blood alcohol content. RCW 46.61.502(1), (2). The determination of that content is by an analysis of breath or blood authorized by RCW 46.61.506. To be valid an analysis of breath or blood "shall have been performed according to methods approved by the state toxicologist". RCW 46.61.506(3). The whole scheme of the crime of driving while intoxicated has withstood numerous challenges. State v. Franco, 96 Wn.2d 816, 639 P.2d 1320 (1982); State v. Baker, 56 Wn.2d 846, 355 P.2d 806 (1960); State v. Erdman, 64 Wn.2d 286, 391 P.2d 518 (1964); State v. Canaday, 90 Wn.2d 808, 585 P.2d 1185 (1978). Particularly instructive on a constitutional challenge is State v. Melcher, 33 Wn. App. 357, 655 P.2d 1169 (1982).
We note that this case is not a contested case within the purview of the administrative procedure act, RCW 34.04, but that the statutory procedures therein provide an analogous methodology of review. In any event, the courts have inherent power to review an administrative action to assure that it was not arbitrary and capricious. Pierce Cy. Sheriff v. Civil Serv. Comm'n, 98 Wn.2d 690, 694, 658 P.2d 648 (1983).
Finally, on review of the decision of the district court, the superior court determines whether that court has committed any error of law. RALJ 9.1(a). Consistent with any *830appellate review of factual findings of a trial court, however, the reviewing court
shall accept those factual determinations [of the district court] supported by substantial evidence in the record (1) which were expressly made by the court of limited jurisdiction, or (2) that may reasonably be inferred from the judgment of the court of limited jurisdiction.
RALJ 9.1(b).
The key issue is whether the state toxicologist's approval by regulation of the DataMaster was arbitrary and capricious. We first need some standard by which to determine whether this action is arbitrary and capricious. The oft-repeated definition is ""'willful and unreasoning action in disregard of facts and circumstances.""' UPS, Inc. v. Department of Rev., 102 Wn.2d 355, 365, 687 P.2d 186 (1984), quoting Skagit Cy. v. Department of Ecology, 93 Wn.2d 742, 749, 613 P.2d 115 (1980). That phrase is traceable to Sweitzer v. Industrial Ins. Comm'n, 116 Wash. 398, 401, 199 P. 724 (1921). The Sweitzer discussion is more expansive than the generally stated rule. The relevant language is:
The most that can be said of their [the agency] action, even from the respondent's point of view, is that they erred in judgment. But this is not arbitrary or capricious action. These terms, when used in this connection, must mean wilful and unreasoning action, action without consideration and in disregard of the facts and circumstances of the case. Action is not arbitrary or capricious when exercised honestly and upon due consideration where there is room for two opinions, however much it may be believed that an erroneous conclusion was reached.
Sweitzer, at 401.
From this language several principles are distilled. First, an error in judgment is not arbitrary and capricious. A judicial conclusion that the administrative decision was erroneous is not sufficient. Second, the action essentially must be in disregard of the facts and circumstances *831involved. Third, the court necessarily looks to the facts before the administrative agency.
From these generalities we must determine whether our review is an examination of a question of law or a question of fact. If we are reviewing a question of fact, our duty is simply to accept factual determinations made by the trial court which are supported by substantial evidence. RALJ 9.1(b).
The defendant contends that appellate review is of a question of law and therefore de novo. We disagree. Judicial review of administrative action may indeed pose a question of law. For example, an agency rule must stem from an express or necessarily implied statutory grant of authority, i.e., delegation. The court reviews such rules to ascertain statutory authority and a reasonable consistency with the statute being implemented. Green River Comm'ty College v. Higher Educ. Personnel Bd., 95 Wn.2d 108, 112, 622 P.2d 826, adhered to and modified, 95 Wn.2d 962, 633 P.2d 1324 (1980). Generally, such questions will be a matter of law.
Here, however, we are concerned with the actions of the state toxicologist. We are not determining whether he acted beyond his authority, but rather whether he acted in disregard of the facts and circumstances before him. The very nature of the inquiry is what the toxicologist did, what facts he relied upon, whether he acted without any rational relation to the facts before him. Conclusions about the action of the toxicologist are based upon what he did or did not do. These are events, occurrences, realities as to what took place. They are facts. Grimwood v. University of Puget Sound, Inc., 110 Wn.2d 355, 753 P.2d 517 (1988).
To review the act of the toxicologist de novo would tend to substitute judicial judgment for an administrative decision. That we are not permitted to do. Indeed, when the rule (which we deem the toxicologist's action to be) is authorized and consistent with the authorizing statute, the *832rule is presumed valid; the one attacking it has the burden of overcoming that presumption. Weyerhaeuser Co. v. Department of Ecology, 86 Wn.2d 310, 314, 545 P.2d 5 (1976). Historically, in similar contexts, this court has emphasized its review of the record to judge conduct alleged to be arbitrary and capricious. See, e.g., Morgan v. Department of Social Sec., 14 Wn.2d 156, 185, 127 P.2d 686 (1942); Robinson v. Olzendam, 38 Wn.2d 30, 37, 227 P.2d 732 (1951).
In approaching this review, we are favorably impressed by the remarks in Natural Resources Defense Coun., Inc. v. SEC, 606 F.2d 1031, 1050 (D.C. Cir. 1979):
In short, the concept of "arbitrary and capricious" review defies generalized application and demands, instead, close attention to the nature of the particular problem faced by the agency. The stringency of our review, in a given case, depends upon analysis of a number of factors, including the intent of Congress, as expressed in the relevant statutes, particularly the agency's enabling statute; the needs, expertise, and impartiality of the agency as regards the issue presented; and the ability of the court effectively to evaluate the questions posed. Only through such a flexible approach can we review the multifarious types of agency actions as responsible participants in an enterprise of practical governance.
(Footnotes omitted.)
Thus, as we turn to the record here, we are mindful that it is not our function to substitute our judgment for that of the state toxicologist, nor was such the function of the trial judge. An unwise or even erroneous decision arrived at pursuant to the legislative duty delegated, upon facts which motivated a rational decision, is not arbitrary and capricious. That the toxicologist might have used a methodology more precise or might have used a different procedure of evaluation reflects upon his administrative judgment, but does not make his action arbitrary and capricious.
Before embarking upon a review of the record to determine whether substantial evidence supports the stated or reasonably inferred factual determinations of the District *833Court, we put in context the decision of Judge Rindal. He heard testimony and arguments for 2 full days, the last day's hearing ending at 10:15 p.m. He issued a 33-page opinion which demonstrates a thorough understanding of the scientific principles involved in the design and operation of the DataMaster. The evidence before Judge Rindal comprises 600 pages of testimony and 41 exhibits. The opinion accurately summarizes and rejects the six grounds raised by the defendants' motion to suppress.
In applying our narrow and limited scope of review to the decision of the District Court, we must remember that the Legislature has mandated that the analysis of breath or blood is valid if it is performed "according to methods approved by the state toxicologist”. RCW 46.61-.506(3). The ultimate concern of the judiciary is that the methods approved result in an accurate test, competently administered, so that a defendant is assured that the test results do in fact reflect a reliable and accurate measure of his or her breath content. We are satisfied that this ultimate goal is reached here as reflected by Judge Rindal's conclusion:
In the instant case, although there was substantial and persuasive evidence that the rules, at the time of adoption were without scientific basis, no witness testified that the DataMasters in the field were not producing accurate and precise results and to the contrary, Sergeant Gullberg testified that the results of the evaluation and certification tests performed on each instrument before it was placed in the field, convinced him that the test results from those instruments were accurate and precise.
Opinion of Judge Rindal, Bellevue District Court, at 30-31 (June 10, 1986).
In reviewing the record it is helpful to summarize the steps taken which led to approval of the DataMaster:
1. In 1983 an ad hoc committee was formed to evaluate the use of an infrared method of breath testing replacing the "Breathalyzer" previously in use for many years. State v. Baker, 56 Wn.2d 846, 355 P.2d 806 (1960). The committee included the state toxicologist and Sergeant Gullberg, *834the State Patrol administrator of the breath test program, and a qualified expert.
2. Ultimately, infrared devices were submitted by four manufacturers including a BAC Verifier, the manufacturer of which eventually produced the DataMaster.
3. The various machines were tested by the State Patrol from September 1983 to April 1984.
4. In June 1984 bid specifications were prepared to meet criteria set by the toxicologist and State Patrol.
5. Contract awarded to manufacturer which produced the Verifier, ultimately called DataMaster.
6. On March 26, 1985, the toxicologist adopted WAC 448-12-210, which approved the BAC Verifier DataMaster infrared breath test instrument. WAC 448-12-220 through 448-12-340 defined the test to be used, methodology, and operator's qualifications.
7. The DataMaster was tested by the State Patrol from April 1985 to January 1986. It was not then in use in the field, at least until December 1985.
8. The toxicologist adopted an emergency rule on December 16, 1985, the main purpose of which was to provide for a more accurate test of the breath sample. WAC 448-12-210 through 448-12-340.
9. The toxicologist adopted nonemergency rules to amend March 1985 rules, again approving the DataMaster device, establishing test definition, procedures, etc.
10. Defendant was cited on March 8,1986.
The main thrust of the defendant's challenge is that the DataMaster in use in the field, and used on this defendant, did not exist when the March 1985 WAC approval was made. Second, the defendant points to 17 changes from the test machine and the DataMaster in use. The Superior Court relied upon these two points to reverse the District Court. These challenges have superficial attraction, but close examination of the procedures employed dispels their appeal.
*835The scientific principles of infrared spectroscopy upon which these machines operate are established and accepted. The four devices submitted for testing, including the BAC Verifier, were based upon those principles.
The toxicologist created the written testing protocol to be used in the testing and evaluation of the four machines. The toxicologist reviewed and relied upon the testing and evaluation data before approving the DataMaster. Blood/ breath correlations were done prior to approval. The toxicologist's opinion was that all four machines produced satisfactory results.
Following substantial testing of the four machines, bid specifications were prepared and reviewed by the toxicologist. While none of the tested machines would have met perfectly the bid specifications, the bid specifications set the criteria for machines to be furnished, all relating to an analysis and performance based upon the same scientific principles. The toxicologist considered the DataMaster to be the same type machine despite some differences in components, but which met the bid specifications. The differences between the Verifier and the DataMaster were not of importance to the toxicologist because both were still measuring alcohol in the same way.
Highly significant and compelling is the toxicologist's instruction that before "any instrument [DataMaster] is placed into the service [in the field for actual use], they have to be reevaluated, certified and blood/breath correlations repeated.'' Transcript of Proceedings, at 184. Further, "every instrument [DataMaster] before it's placed in the field should be evaluated for precision, accuracy — for precision and accuracy." Transcript of Proceedings, at 186.
Before December 1985 WAC approval of the DataMaster, additional tests were done with results comparable to the original BAC Verifier. The toxicologist concluded, before the DataMaster was ever put to actual use, that the instruments were accurate and precise. Transcript of Proceedings, at 216.
*836Sergeant Gullberg conducted or supervised most of the tests on the initial machines and the DataMaster. He testified that in addition to tests conducted at the direction of the toxicologist, the National Highway Traffic Safety Administration, in July 1985, tested the DataMaster. The nature of the tests and the results were published. The DataMaster met the federal standards and was placed on an approved products list. That evaluation was relied upon in part by Sergeant Gullberg in his evaluation of the Data-Master. It was Sergeant Gullberg s opinion that the Data-Master was a reliable, accurate and precise instrument for its intended purpose. Report of Proceedings, at 528.
Much emphasis is placed upon changes between the test machine and the one in use. Such changes could be significant, but no testimonial record was made of the significance of those changes. Defendant notes the existence of 17 changes, as did Judge Rindal, but defendant offers no evidence of the significance of those changes. For example, there is a difference in circuitry and a difference in radio frequency interference antennas. These changes do not overcome the positive expert opinions as to the accurate and precise performance of the DataMaster which tested defendant.
It may well have been premature for the toxicologist to issue his March 1985 approval before the ultimate machine was at hand. However, before the DataMaster was applied to defendant Ford, it had been tested, retested and certified. The toxicologist, to whose satisfaction the device must perform according to legislative mandate, issued a new WAC again approving the DataMaster. This occurred more than 3 months before defendant was cited.
In summary, there is nothing in the record to suggest that the DataMaster produced an inaccurate result when its test was administered to defendant Ford. There was no witness who testified that the DataMasters in actual use, after final approval by the toxicologist, were not producing accurate and precise results.
*837There was substantial evidence to support the factual determinations made by the District Court; the judge committed no error of law. Therefore, the Superior Court is reversed.
Pearson, C.J., Utter, Dolliver, Dore, Callow, and Durham, JJ., and Cunningham, J. Pro Tern., concur.
Pearson, C.J. — By order of the court, SAR 14 has been suspended for purposes of this appeal and the dissent to this opinion will be filed at a later date.
[En Banc. June 30, 1988.]