OPINION

Early in the morning of May 31, 1982, Marcelo Quinto, Jr., was arrested for driving a motor vehicle while under the influence of intoxicants (DWI) in violation of City and Borough of Juneau Municipal Code § 72.10.010.1

Quinto was observed driving his Ford Bronco erratically near Whittier Street in downtown Juneau by a Juneau police officer, who communicated Quinto’s location, license plate,number, and make of car to another officer, Corporal Karl Lewkowski. Lewkowski drove to the area and saw a “Bronco-type vehicle” with its brake lights on at the top of a ramp leading to the Prospector Hotel; no other vehicles were in sight.

Lewkowski drove up the ramp and verified the license plate number on the Bronco. He then approached Quinto, who was behind the wheel; as he walked towards the Bronco, Lewkowski activated a small tape recorder that was attached to his belt. When he got to the Bronco, Lewkowski *632noticed that Quinto had bloodshot eyes and smelled of alcohol. Lewkowski concluded that Quinto was intoxicated and asked him to perform field sobriety tests. After performing poorly on the sobriety tests, Quinto was arrested. All pre-arrest communications between Lewkowski and Quinto were recorded without Quinto’s knowledge on Lewkowski’s tape recorder. Quinto later repeated the sobriety tests on videotape at the Juneau Police Department, but he refused to submit to a breathalyzer examination.

On appeal, Quinto asserts that the trial court committed error by refusing to allow expert testimony concerning Quinto’s blood alcohol level at the time of his arrest, by excluding testimony concerning Quinto’s reputation for sobriety in the community, and by denying a motion to suppress the tape recording that was made without Quinto’s permission prior to his arrest. We reverse.

At trial, Quinto attempted to introduce expert testimony by Dr. Gary Christian concerning Quinto’s blood alcohol level at the time of his arrest. The trial court permitted Quinto to voir dire Dr. Christian for the purpose of establishing Dr. Christian’s qualifications. Dr. Christian stated that he was a tenured professor of chemistry at the University of Washington and had studied the breathalyzer machine, as well as several alternative methods of establishing blood alcohol levels. If allowed to testify, Dr. Christian would have used a method known as “Widmark’s formula” to establish-Quin-to’s blood alcohol level at the time of his arrest. According to Dr. Christian, an accurate measurement of a person’s blood alcohol level can be obtained through use of Widmark’s formula when the amount and weight of the alcohol consumed, the weight of the consumer, and the time sequence of the consumption are all known. For his computations, Dr. Christian apparently planned to rely upon Quinto’s testimony with respect to the nature, amount, and timing of his drinking prior to arrest. Using these figures, Dr. Christian would presumably have testified that Quinto’s blood alcohol level was relatively low at the time of his arrest.

Upon completion of voir dire, defense counsel stated that Dr. Christian’s testimony was offered to show that Quinto was not under the influence of intoxicating beverages when driving and to corroborate Quin-to’s testimony that he was not intoxicated. The prosecution objected to admission of Dr. Christian’s testimony, arguing that evidence concerning Quinto’s blood alcohol level was irrelevant to the charge of DWI; the prosecutor maintained that Quinto’s blood alcohol level would have been relevant only if he had been charged with the separate offense of driving with an excessive blood or breath level, in violation of CCBJ 72.10.-0112.

The trial court concluded that in the absence of a breathalyzer result for Quinto, testimony about Quinto’s blood alcohol level established by a method other than the breathalyzer was irrelevant. The trial court also found that Dr. Christian’s testimony would unnecessarily distract and confuse the jury.

On appeal, Quinto argues that the trial court abused its discretion by excluding Dr. Christian’s testimony. Quinto asserts that Dr. Christian’s testimony would have been relevant, under Evidence Rule 401,3 to the question of whether Quinto was intoxicated. Quinto also disputes the accuracy of the trial court’s finding that exclusion of this testimony was necessary because of its potential for jury confusion.

*633We conclude that Quinto’s arguments concerning the trial court’s exclusion of Dr. Christian’s testimony are meritorious. In Denison v. State, 630 P.2d 1001 (Alaska App. 1981), we considered a factual situation which was the converse of that presented in this case. Denison was charged with driving with a blood alcohol level of .10% or greater. At trial, Denison attempted to offer into evidence a videotape made by police, after her arrest, as circumstantial proof that her breathalyzer result was incorrect. Denison also attempted to present witnesses to testify about her apparent sobriety and the amount of alcohol she consumed before her arrest. The district court refused to permit Denison to offer either the videotape or hdl witnesses, ruling that nontechnical evidence offered to prove the inaccuracy of a breathalyzer result was inadmissible unless technical evidence of inaccuracy was also presented. In reversing Denison’s conviction, we found that the evidence offered was relevant circumstantial evidence of the possibility that Denison’s breathalyzer result was erroneous, and we noted the preference for admission of relevant evidence that is established under Evidence Rule 402.4 We also concluded that exclusion of Denison’s testimony was not justified, under Evidence Rule 403,5 by the danger that the jury would be misled: “[W]e fail to see how a jury, properly instructed as to the purpose for which such evidence was offered, could have been prejudiced, confused, or misled.” Denison v. Anchorage, 630 P.2d at 1003. We subsequently relied upon Denison in Byrne v. State, 654 P.2d 795 (Alaska App. 1982). Byrne was convicted of driving with a blood alcohol level of .10% or more, in violation of AS 28.35.030; on appeal Byrne contended that the trial court erred by admitting testimony that he was driving erratically and appeared intoxicated at the time of his arrest. We found this point to be controlled by Denison, and we stated:

We are satisfied that there is a sufficient nexus between erratic driving and other indicia of intoxication and an elevated blood alcohol level that such evidence would tend to corroborate a breathalyzer reading showing an elevated blood alcohol level. The evidence was therefore relevant to this prosecution.

Byrne argues, nevertheless, that the probative value of the evidence was so weak and its prejudice so obvious that the trial court abused it discretion in failing to exclude the evidence under Alaska Rule of Evidence 403. We disagree. We believe the evidence in question highly probative of the accuracy of the breathalyzer reading. While we agree with Byrne that evidence of his appearance and erratic driving tended to portray him to the jury as a drunk driver, we believe that such a portrayal was a necessary corollary of the prosecution in question. It certainly is not the kind of prejudice that Evidence Rule 403 seeks to prevent, i.e., a decision by a jury on grounds totally unrelated to the elements of the crime in question.

Byrne v. State, 654 P.2d at 796.

In the present case, the City of Juneau argues that Denison and Byrne are factually distinguishable because they involved charges of driving with a blood alcohol level of .10% or greater and because the challenged testimony was that of witnesses who had direct knowledge of the defendants’ appearance and alcohol consumption prior to arrest. The City of Juneau also argues that, even if relevant, Dr. Christian’s testimony was properly excluded because of its potential for distracting and misleading the jury.

*634We disagree with the narrow reading of 'Denison and Byrne proposed by the City of Juneau. As we noted in those cases, the relationship between high blood alcohol levels and intoxication is a matter of common sense and understanding. In Denison and Byrne, testimony and evidence concerning the defendants’ apparent condition prior to arrest was offered as circumstantial proof of the accuracy or inaccuracy of blood alcohol tests. Here, a blood alcohol computation was offered as evidence that the defendant’s actual condition was not impaired by consumption of alcohol. The City of Juneau has failed to demonstrate why evidence of a defendant’s blood alcohol level is less relevant to establish lack of actual impairment than evidence of apparent sobriety or lack of sobriety is relevant to confirm or impeach the results of blood alcohol tests. We find that Dr. Christian’s testimony was circumstantial evidence that would have been relevant to rebut the accuracy of Corporal Lewkowski’s testimony concerning Quinto’s apparent intoxication.

Similarly, we disagree with the City of Juneau’s assertion that this evidence could properly be excluded under Evidence Rule 403. The potential for confusion of the jury would have been minimal if the court had instructed the jury concerning the proper use of circumstantial evidence. While the evidence upon which Dr. Christian would have based his testimony — Quin-to’s own testimony as to the amount of alcohol he had consumed — might be subject to a healthy measure of skepticism in light of Quinto’s refusal to take a breathalyzer test, this is plainly a matter of credibility that should properly have been decided by the jury. The issue is certainly not so inherently confusing or obscure as to justify excluding Dr. Christian’s testimony out of fear that the jury might be misled. We hold that exclusion of Dr. Christian’s testimony was error. In the context of the evidence presented at trial, we cannot say that the error was harmless. Accordingly, Quinto’s conviction must be reversed.6

While we have concluded that Quin-to’s conviction must be reversed, we must also consider the merits of Quinto’s other arguments, since they raise issues that are likely to recur upon retrial. Quinto contends that the court erred in refusing to permit testimony by Kyle Reiger, Quinto’s co-worker, that Quinto had a reputation in the community for being “a cautious, sober individual.” The trial court did allow testimony by Mr. Reiger that Quinto had a reputation for truthfulness and veracity in the community. Quinto argues that the testimony as to sobriety was properly admissible under Evidence Rule 404(a)(1), which states:

(a) Character Evidence Generally. Evidence of a person’s character or a trait of his character is not admissible for the purpose of proving that he acted in conformity therewith on a particular occasion, except:

(1) Character of Accused. Evidence of a relevant trait of his character offered by an accused, or by the prosecution to rebut the same. . . .

We fail to see why testimony about Quin-to’s reputation for sobriety in the community would not tend to establish a relevant trait of his character, within the meaning of Evidence Rule 404(a); certainly, a reputation for sobriety is as relevant to a charge of DWI as is a reputation for truthfulness or veracity. We hold that Quinto should not have been precluded from offering this testimony at trial.

*635Finally, Quinto argues that the trial court erred in not suppressing the recording Lewkowski made of his conversation with Quinto during administration of the field sobriety tests prior to Quinto’s arrest. Quinto contends that reversal of the trial court’s decision is mandated by State v. Glass, 583 P.2d 872 (Alaska 1978). In Glass, a police informant carried a concealed radio transmitter into the home of Glass, from whom the informant had arranged to purchase heroin. The conversation between Glass and the informant was electronically monitored and recorded by police officers located outside the home. The supreme court concluded that article 1, section 22 of the Alaska Constitution7 prohibits the surreptitious electronic monitoring of conversations upon the consent of only one participant. 583 P.2d at 879-81. The court relied, in part, on the two pronged standard for determining privacy that was articulated by Justice Harlan in Katz v. United States, 389 U.S. 347, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967): (1) Whether the defendant exhibited an actual expectation of privacy; and (2) whether that expectation was one that society is prepared to recognize as reasonable. See id. at 361, 88 S.Ct. at 516-17, 19 L.Ed.2d at 588 (Harlan, J., concurring).

Quinto argues that Glass must be read to prohibit the warrantless recording of his pre-arrest conversation with Corporal Lewkowski. In response, the City of Juneau asserts that Glass applies only to situations in which warrantless, nonconsensual electronic surveillance is conducted by use of an informant or an undercover officer. The city maintains that because Corporal Lewkowski was in uniform and involved in a routine investigatory stop when the recording was made, any expectations on the part of Quinto that his conversation was private was unrealistic. The city also contends that Palmer v. State, 604 P.2d 1106 (Alaska 1979), supports its restrictive reading of Glass.

In Palmer, the defendant relied on Glass to argue that the trial court erred in refusing to suppress a videotape made without his consent after he was arrested for DWI. The supreme court disagreed and found Glass to be distinguishable:

We think the situation in the case at bar, however, is readily distinguishable. When the videotape recording was made, Palmer was already under arrest. After being transported to police headquarters, he was asked to submit to a breathalyzer examination and to perform a number of sobriety tests. Assuming, arguendo, that he had any actual or subjective expectation at that point that his actions would not be recorded, we are convinced that that expectation is not one that society is prepared to recognize as reasonable. Accordingly, we hold that there was no violation of the right of privacy guaranteed to Palmer by article 1, section 22, of the state constitution.

Palmer v. State, 604 P.2d at 1108 (emphasis in original).

We think that the Glass holding cannot be restricted to situations involving undercover police informants. One of the basic premises of the supreme court’s opinion in Glass is that warrantless monitoring of conversations has a definite chilling effect upon freedom of expression when only one of the participants has consented. State v. Glass, 583 P.2d at 877-78. Because of the factual setting in which the case arose, much of the discussion in Glass is couched in terms of police informers and undercover agents. However, there is also broad language implying that the warrant requirement is applicable to the situation posed by the present case:

[W]e believe that Alaska’s privacy amendment prohibits the secret electronic monitoring of conversations upon the. mere consent of a participant ... We conclude that the expectation that one’s *636conversations will not be secretly recorded or broadcast should be recognized as reasonable.

State v. Glass, 583 P.2d at 879-80.

We believe that Palmer actually lends support for Quinto’s interpretation of Glass. In Palmer, the supreme court was afforded an opportunity to limit Glass by rejecting the reasonableness of an individual’s expectation of privacy in his communications with uniformed police officers. Instead, the court focused on Palmer’s arrest, detention, and presence in the police station as the significant factors justifying the conclusion that there could be no reasonable expectation of privacy. 604 P.2d at 1108. We conclude that the warrant requirement of Glass must be read to include situations involving routine nonconsensual recording of pre-arrest conversations between citizens and uniformed officers.8 We therefore hold that the trial court committed error in denying Quinto’s motion to suppress the tape recording made by Corporal Lewkowski; upon retrial, the recording should not be admitted as evidence.

Quinto’s conviction is REVERSED and the case is REMANDED for further proceedings consistent with this opinion.

concurring in part and dissenting in part.

I agree with the court’s disposition of all issues except the suppression of the recording of Quinto’s remarks to the arresting officer. I cannot see how anyone could have an expectation of privacy in the contents of a conversation with a uniformed police officer. Palmer v. State, 604 P.2d 1106, 1108 (Alaska 1979), and see People v. Crowson, 33 Cal.3d 623, 190 Cal.Rptr. 165, 660 P.2d 389 (Cal. 1983) (defendant had no reasonable expectation of privacy while seated in back seat of police car with accomplice so that surreptitious tape-recording of their conversation was admissible in evidence).