The issue before us is whether there was probable cause for the issuance of a search warrant. The warrant was issued by a Judge of the Court of Special Sessions of the City of Yonkers upon the application and affidavit of a police officer. County Court granted defendant’s motion to suppress evidence of his participation in a gambling operation obtained by way of a search of his person pursuant to the warrant. With two Justices dissenting, the Appellate Division reversed. We hold that there was a lack of probable cause and, accordingly, reverse.
The information from a confidential informant described in the affidavit of the police officer did not satisfy the requirements of Aguilar v Texas (378 US 108). Aguilar established this two-prong standard: the Magistrate issuing the warrant must be informed of (1) some of the underlying circumstances from which the informant concluded that illegal activities were taking place, and (2) some of the underlying circumstances from which the officer concluded that the informant was credible or reliable (378 US, at p 114). Although the latter prong was satisfied, the former, sometimes called the "basis of knowledge test” (People v Hanlon, 36 NY2d 549, 556), was not. This was not a situation where the affidavit merely failed *132to disclose certain details. For example, had the affidavit disclosed the underlying circumstances from which the informant concluded that illegal activities were taking place, but failed to specify dates of his observation, detailed confirmatory observations of the police would be sufficient to verify the trustworthiness of his information (People v Brandon, 38 NY2d 814). Here, the affidavit failed to disclose any of the underlying circumstances from which the informant concluded that defendant was engaged in gambling activities. The purpose of the Aguilar requirement is to allow a "neutral and detached magistrate” to determine independently whether the informant was justified in his conclusion (378 US, at p 115; see Giordenello v United States, 357 US 480, 486). Absent any recitation of the underlying circumstances, it was impossible for the Magistrate to perform this function which this court has suggested is "the primary basis for the Aguilar holding” (People v Brown, 40 NY2d 183, 186; see, also, People v Hendricks, 25 NY2d 129; cf. People v Montague, 19 NY2d 121).
Since the affidavit did not meet the standards established in Aguilar and thus the Judge issuing the warrant could not, within these standards, have determined that there was probable cause based on the "tip” supplied by the informant, it is necessary to consider whether the other information supplied in the affidavit was sufficient to justify a finding of probable cause. This information was: first, that the police were aware that defendant was a known policy runner and had been arrested for gambling activities only 20 days prior to the signing of the warrant and, second, that the police conducted a surveillance of defendant to confirm the informant’s tip.
With respect to the surveillance of defendant, the police officer’s affidavit recites:
"Your deponent states that on Monday May 21, 1973 between the hours of 11:30 A.M. to 12:30 P.M. a surveillance was conducted of 8-10 School St. Yonkers, N.Y. At about 12:10 P.M. Stephen Wirchansky was observed driving a vehicle south on School St. slowing said vehicle down in the area of 8-10 School St. and looking in the area of that said address; then continuing south. A few moments later said vehicle and operator again came into School St. and parked his vehicle approximately 100 feet south of 8-10 School St. on the east side of the street. Said Wirchansky left vehicle walked to and entered the hallway of 8-10 School St. and remaining therein *133for a few minutes exiting same proceeded to his auto and drove off.
"Your deponent further states that on May 22, 23, 24, 1973 (Tues, Wed. Thurs.) between aforementioned times Stephen Wirchansky was observed continuing the same procedures as mentioned on Monday May 21, 1973.”
The question is whether these observations indicate criminal activities on the part of defendant. A similar description of the movements of an individual engaged in gambling activities was set forth in the affidavit considered in Spinelli v United States (393 US 410). There, the FBI affidavit stated that information had been given by a confidential informant but failed, as in the instant case, to disclose any of the underlying circumstances from which the informant concluded that gambling was taking place. There, as here, the Government sought to corroborate the informant’s tip with its observations of the suspect’s conduct. As to these observations, the court concluded: "Spinelli’s travels to and from the apartment building and his entry into a particular apartment on one occasion could hardly be taken as bespeaking gambling activity” (393 US, at p 414). Nevertheless, although the observations in Spinelli were held insufficient to corroborate the informant’s tip, the instant matter is somewhat different because the police claim that appellant’s movements in fact "bespeak” gambling activities.
The affidavit under review contains this statement regarding the informant’s tip: "Your deponent also states that information [was] received from a confidential informant that gambling paraphernalia commonly associated with a bookmaking and policy scheme, are being left in a common hallway of 8-10 School St. mainly a mailbox, which is picked up by Stephen Wirchansky each day.”
Although this information was quite specific, since the affidavit does not disclose how the informant came to know these facts, the Aguilar test was nonetheless not satisfied. To measure up to Aguilar, the affidavit should have described either that the informant saw defendant make a pickup (see People v Hendricks, 25 NY2d 129, 136, supra) or, if the informer did not see defendant, the sources from which the informer learned of defendant’s activities (People v Sutton, 32 NY2d 923, cert den 415 US 983). In this respect, similar specific information from an informer, i.e., specific telephone numbers from which wagering information was disseminated, *134was rejected in Spinelli because: "This meager report could easily have been obtained from an offhand remark heard at a neighborhood bar” (393 US, at p 417). In the instant matter, the informer could have overheard that defendant was making a pickup from a mailbox at a specific address—a rumor such as that would not amount to probable cause for a search of defendant.
In Spinelli, the FBI’s affidavit stated that an informant had advised it that "Spinelli is operating a handbook and accepting wagers” (393 US, at p 422). By contrast, the police affidavit here concluded: "It also should be noted that it is your deponent’s opinion based upon numerous gambling investigations in which your deponent has been involved over a period of the last two (2) years that Stephen Wirchansky’s activities are that of a pick-up man in a policy operation. This is based upon observations of Wirchansky’s activities and especially the times in which this activity takes place.” Thus, unlike Spinelli where the observations of daily travels did not indicate any particular activity other than to place the suspect at the location of the gambling operation, the activity described here is, according to an experienced police officer, that of a pick-up man in a policy operation.
There have been instances where certain conduct which, although innocent-seeming to the untrained eye, may indicate illegal conduct to a trained police officer (People v Valentine, 17 NY2d 128, 132; see People v Castro, 29 NY2d 324, 327). Those instances are, however, unusual because the expertise of the police officer is generally not enough to justify a search of one engaged in seemingly innocent conduct (see People v Corrado, 22 NY2d 308; cf. People v Davis, 36 NY2d 280; People v Brown, 32 NY2d 172, 174). The principle is aptly expressed as follows: "The logical and practical problem is that even accepting ungrudgingly, as one should, the police officer’s expertness in detecting a pattern of conduct characteristic of a particular criminal activity, the detected pattern, being only the superficial part of a sequence, does not provide probable cause for arrest if the same sketchy pattern occurs just as frequently or even more frequently in innocent transactions. The point is that the pattern is equivocal and is neither uniquely nor generally associated with criminal conduct, and unless it is there is no probable cause.” (People v Brown, 24 NY2d 421, 424.)
Even paying great respect to the specialized knowledge of *135the police officer in this situation, we cannot say that the conduct described in the affidavit was susceptible of only one interpretation—criminal activity. Although defendant’s conduct was suspicious, he may have been entering the apartment building for innocent purposes totally unrelated to the gambling operation, and thus his conduct would not sustain a finding of probable cause. In so concluding, we distinguish those cases cited by the People where the conduct observed by the police was not consistent with innocent conduct (see People v Valentine, 17 NY2d 128, supra; cf. People v Meyers, 38 AD2d 484).
The only other information provided by the police in its affidavit was that defendant was known to the police and had been arrested for gambling activities. Although criminal reputation is relevant when supported by other information (see United States v Harris, 403 US 573, 582-583), it is not enough alone to save the warrant in this case. The holding in Harris did not nullify the principle expressed in Spinelli that a suspect’s reputation may not be used "to give additional weight to allegations that would otherwise be' insufficient” (Spinelli v United States, 393 US 410, 418-419, supra).
In holding that this warrant was not issued upon probable cause, we distinguish diverse patterns. If a policeman relies on an informant to show probable cause for a search, the affidavit, to satisfy the Aguilar standards, must disclose the underlying facts from which the informant concluded that illegal activity is taking place (People v Hendricks, 25 NY2d 129, supra; accord People v Brown, 40 NY2d 183, 186, supra). In many instances where the underlying circumstances disclosed suffer from a lack of detail, the facts may be bolstered by police observations (see, e.g., People v Brandon, 38 NY2d 814, supra; People v Slaughter, 37 NY2d 596, 599; People v Hanlon, 36 NY2d 549, supra). Where, however, the informant’s tip does not disclose any of the circumstances of the criminal activity, the police are relying on the informant only to alert them to the person who is engaged in the activity, after which the police act on their own observations (see People v Castro, 29 NY2d 324, 327, supra; People v Meyers, 38 AD2d 484, supra [both involving warrantless searches]). In these instances, which include the matter before us, the police observations must do more than merely bolster the informant’s tip, since the police themselves must provide the underlying circumstances from which the Magistrate can determine *136whether they are justified in concluding that illegal activity is taking place.
Testing this affidavit in this fashion, we are presented these facts: a tip from a reliable informer; police observations of conduct which is as consistent with innocent as it is with criminal activity; and defendant’s criminal reputation with respect to the particular crime of which he is suspected. Surely, if merely one of these three items was present there would not be probable cause. Yet, in a given case the totality of circumstances might support a finding of probable cause (see United States v Harris, supra). Here, however, the crucial element is the police observations. Since the observed activities are susceptible of an interpretation of innocent activity, the informant’s tip and the defendant’s reputation should not be allowed to elevate these activities to the level of probable cause. To do so would be to declare that every time the police receive an informant’s tip that a ’’known criminal” is going to a particular location for criminal purposes, police observations of "suspicious conduct” of the suspect at that location give rise to probable cause for a search.
The dissent urges that a Magistrate examining the affidavit in a "commonsense” and "realistic fashion” could determine that there was probable cause for the search in this case (dissent, at p 137). Yet, the very suggestion that a commonsense approach be used manifests the inability to satisfy the appropriate legal criteria—as firmly established by a long line of decisions both of this court and of the Supreme Court of the United States. In this connection, it is salutory that the principle of stare decisis be followed and that there be stability to the law.
If the conduct of defendant were as suggestive of criminal activity (compare People v Seney, 34 NY2d 817) as the dissent implies, it would be unnecessary for us to consider whether the informant’s tip and the defendant’s reputation, considered in connection with police observations, give rise to probable cause. All we have here is "highly suspicious” conduct and such suspicions do not amount to probable cause. That the police went before a Magistrate and requested a warrant, rather than conducting a warrantless search, should not alter our conclusions. There is no question of the "bona ñdes of the police” (see dissent, at p 139). We are not unmindful of the preference for search warrants and we do not mean to suggest by our holding any lack of confidence in the Magistrate in *137fulfilling his duty to scrutinize and weigh the information presented to him. We simply hold that when the police rely on an informant who has failed to disclose any of the circumstances from which he concluded that criminal activities are taking place, more is necessary than the defendant’s reputation and the police observations in this case to support a finding of probable cause.
Accordingly, the order of the Appellate Division should be reversed, and the motion to suppress should be granted.