*140OPINION OF THE COURT
Levine, J.These appeals arise out of motions to suppress evidence introduced against occupants of taxicabs that were stopped by police officers assigned to a Taxi-Livery Task Force in New York City. The procedure of the police was essentially the same in both cases. The officers were in plain clothes, using unmarked police cars, on a roving patrol. They selected a location in an area having a high incidence of taxi robberies. In the absence of probable cause, suspicious behavior or any illegal activity on the part of the driver or passengers, they pulled over a predetermined percentage of livery vehicles. The purpose of the stops was to provide a “safety check,” to give the drivers crime prevention information and make an assessment of whether the driver may have been in danger. For the reasons that follow, we conclude that the stops in both of these cases were unreasonable seizures violating the Fourth Amendment of the US Constitution.
I.
The material facts in these cases are not in dispute. In Muhammad F., two police officers wearing civilian clothes and driving in an unmarked police car, pulled over a cab in which respondent was a passenger and arrested him after he was found to be in possession of a controlled substance. Respondent was subsequently adjudicated a juvenile delinquent.
According to the suppression hearing testimony of the commanding officer of the Taxi-Livery Task Force, the unit was created in 1992 to address the high incidence of violent crimes against cab drivers. The taxi - program targeted certain neighborhoods, especially between 6:00 p.m. and 2:00 a.m. Special attention was given to cabs with passengers. Members of the Task Force would stop a taxi, hand the driver a pamphlet with safety tips and question the cab driver as to his safety, while at the same time watching the reaction of the passengers in the cab. Generally, such a safety check would involve asking the passengers to “step out briefly” or “telling passengers to get out” while the officers searched around and under the seats.
There were no written guidelines as to how the taxi stops were to be carried out. Instead, the officers proceeded on verbal instructions under which the stops were to be done “in a set basis and not just arbitrarily.” Whether the officers stopped every cab, or every second, third or fifth taxi was not based on *141a fixed format, but varied, depending on circumstances seen and acted upon solely by the officers in the field. The number of stops the officers were to make on their tour throughout the night was totally within their discretion. The officers did not keep any written record of stops they had made.
On the night of July 7, 1996, the patrolmen parked their unmarked car just off 145th Street in Manhattan to conduct taxi safety checks. The officers decided to check one in three occupied cabs. The third vehicle the officers saw which they believed to be an occupied taxicab was the non-medallion taxi in which respondent was riding. They turned onto 145th Street to drive up behind the vehicle, put a flashing light on the dashboard and pulled over the vehicle. One officer then approached on foot while shining a flashlight into the car. When he saw that respondent was acting suspiciously, he ordered the passengers out of the car and searched the back of the cab, where he found a paper lunch-style bag containing crack cocaine. Respondent was arrested.
Supreme Court denied Muhammad F.’s motion to suppress the evidence obtained as a result of this stop, and respondent was ultimately adjudicated a juvenile delinquent for committing acts which, if committed by an adult, would constitute criminal possession of a controlled substance. The Appellate Division reversed (255 AD2d 168), holding that the evidence against respondent was the product of an unconstitutional seizure. One Justice dissented. Petitioner presentment agency appealed pursuant to leave granted by the Appellate Division.
II.
In Boswell, defendant was arrested by a plainclothes Sergeant and patrolman belonging to the Taxi-Livery Task Force after their unmarked car stopped a taxicab in which defendant was a passenger. Boswell was also found to be in possession of illegal drugs.
At the suppression hearing, the patrolman testified that the Taxi-Livery Task Force established locations to stop cabs and hand out safety tips and other literature to the driver. The officers usually stopped every third cab, but there were no written procedures. The Sergeant supervisor determined when to initiate and suspend the operation.
The officers, both in plain clothes, were working a night shift on February 27, 1995, when the Sergeant decided to set up operations on Exterior Avenue in the Bronx. They would stop *142every third livery vehicle, whether it was occupied or not— although there was no testimony as to how they could determine that a car with no passengers in the back seat was a non-medallion taxicab.
The two officers did not stop the first two passing cars they believed to be cabs. When a third passed by them, the officers followed it and pulled it over. After the taxi had stopped, the testifying officer shined his flashlight at defendant, noticed he was attempting to conceal a plastic bag and asked him to exit the cab. The bag contained packets of crack cocaine, and defendant was charged with criminal possession of a controlled substance in the third degree.
Supreme Court suppressed the evidence obtained as a result of the stop. The Appellate Division reversed (255 AD2d 173), but one Justice dissented and granted defendant permission to appeal to this Court.
III.
A brief, suspicionless stop of an automobile, while far less intrusive than a traditional arrest, nonetheless qualifies as a seizure (see, Michigan Dept. of State Police v Sitz, 496 US 444, 450; People v Spencer, 84 NY2d 749, 752). The reasonableness of such a seizure “depends ‘on a balance between the public interest and the individual’s right to personal security free from arbitrary interference by law officers’ ” (Brown v Texas, 443 US 47, 50 [emphasis supplied; internal citations omitted]). Determining whether these stops are constitutional requires a weighing of “[1] the gravity of the public concerns served by the seizure, [2] the degree to which the seizure advances the public interest, and [3] the severity of the interference with individual liberty” (id.; at 50-51). A critical requirement for all such seizures relates to the “central concern * * * that an individual’s reasonable expectation of privacy is not subject to arbitrary invasions solely at the unfettered discretion of officers in the field” (id., at 51). Namely, “the seizure must be carried out pursuant to a plan embodying explicit, neutral limitations on the conduct of individual officers” (id.; see also, People v Spencer, supra, at 758-759).
In Michigan Dept. of State Police v Sitz, the Supreme Court stated that the balancing analysis of Brown v Texas and earlier precedents should be utilized in these “cases dealing with police stops of motorists on public highways” (Michigan Dept. of State Police v Sitz, supra, at 450). Employing these standards, in United States v Brignoni-Ponce (422 US 873), the Court had *143previously held that a suspicionless, random “roving-patrol” by Border Patrol officers to conduct a brief review of the citizenship or immigration status of the vehicle’s occupants was unconstitutional. The roving-patrol stop was described as follows: “a fixed checkpoint * * * was closed because of inclement weather, but two officers were observing northbound traffic from a patrol car parked at the side of the highway. The road was dark, and they were using the patrol car’s headlights to illuminate passing cars. They pursued respondent’s car and stopped it” {id., at 874-875). Preventing “the illegal entry of aliens at the Mexican border” was a “valid public interest” {id., at 878-879). The Court held that the stops were unconstitutional, nevertheless, after taking into account “the availability of alternatives to random stops unsupported by reasonable suspicion” {id., at 883, n 8). Significantly, the Court emphasized that the stops would have subjected residents “to potentially unlimited interference with their use of the highways, solely at the discretion of Border Patrol officers” {id., at 882).
By contrast, United States v Martinez-Fuerte (428 US 543) upheld suspicionless stops by the Border Patrol of northbound vehicles at a fixed checkpoint some 60 miles from the Mexican border, on a major artery from the border (id., at 545). At the checkpoint, still without reasonable suspicion, a uniformed agent directed some of the automobiles to a secondary inspection area for further inquiry regarding the citizenship and immigration status of the occupants (id., at 546-547). After noting the substantial public interest furthered by the practice of routine stops at permanent checkpoints near the border (id., at 556), the Court concluded that a reasonable suspicion requirement “would be impractical because the flow of traffic tends to be too heavy to allow * * * particularized study,” and “such a requirement would largely eliminate any deterrent to the conduct of well-disguised smuggling operations” (id., at 557 [emphasis supplied]).
As with the roving-patrol stop, the “objective intrusion — the stop itself, the questioning, and the visual inspection” — was “quite limited” (id., at 557-558 [emphasis supplied]). But “the subjective intrusion — the generating of concern or even fright on the part of lawful travelers — is appreciably less in the case of a checkpoint stop” (id., at 558 [emphasis supplied]). At routine checkpoint stops, motorists can see that other vehicles are being stopped and “are not taken by surprise” (id., at 558-559). The checkpoint, however, was not required to scrutinize every car in the same manner ( {id., at 563). In fact, the “selective *144referrals — rather than questioning the occupants of every car— tend[ed] to advance some Fourth Amendment interests by minimizing the intrusion on the general motoring public” (id, at 560).
In addition, the Court in Martinez-Fuerte reemphasized the decisiveness of limiting the discretion of officers in the field.
“[Checkpoint operations both appear to and actually involve less discretionary enforcement activity. * * * The location of a fixed checkpoint is not chosen by officers in the field, but by officials responsible for making overall decisions as to the effective allocation of limited enforcement resources. * * * And since field officers may stop only those cars passing the checkpoint, there is less room for abusive or harassing stops of individuals than there was in the case of roving-patrol stops. Moreover, a claim that a particular exercise of discretion in locating or operating a checkpoint is unreasonable is subject to post-stop judicial review” (id, at 559).
Subsequently, Delaware v Prouse (440 US 648) held that a police officer may not arbitrarily pull over an automobile merely to check a driver license and registration. “In those situations in which the balancing of interests precludes insistence upon ‘some quantum of individualized suspicion’ * * * other safeguards are generally relied upon to assure that the individual’s reasonable expectation of privacy is not ‘subject to the discretion of the official in the field’” (id, at 654-655). Thus, in assessing the validity of the suspicionless patrol stop of an automobile chosen at random, the Court weighed (1) the State’s “vital interest in ensuring * * * that licensing, registration, and vehicle inspection requirements are being observed” (id, at 658); (2) “whether in the service of these important ends the discretionary spot check [was] a sufficiently productive mechanism to justify the intrusion” (id, at 659); and (3) “the physical and psychological intrusion visited upon the occupants of a vehicle by a random stop to check documents” (id, at 657).
The Court emphasized that “[t]his kind of standardless and unconstrained discretion is the evil the Court has discerned when in previous cases it has insisted that the discretion of the official in the field be circumscribed” (id, at 661). Thus, Delaware v Prouse held that “persons in automobiles on public *145roadways may not for that reason alone have their travel and privacy interfered with at the unbridled discretion of police officers” (id., at 663; see also, People v Spencer, supra, 84 NY2d, at 758-759).
By contrast, suspicionless stops of vehicles to conduct sobriety checks at checkpoints under written guidelines are constitutional, even if the location of the roadblock regularly changes (Michigan Dept. of State Police v Sitz, supra, 496 US 444; People v Scott, 63 NY2d 518). After reaffirming the applicability of Brown v Texas, the Court in Sitz recognized the magnitude of the governmental interest in combating drunken driving (Michigan Dept. of State Police v Sitz, supra, at 450-451; see also, People v John BB., 56 NY2d 482, 488).
Sitz also addressed “ ‘the degree to which the seizure advance [d] the public interest’ ” (Michigan Dept. of State Police v Sitz, supra, at 453). The Court strongly emphasized that in weighing this Brown v Texas factor, a court is not to assess the propriety and the need of the particular means used by law enforcement authorities to advance the public interest. The “effectiveness” part of the test “was not meant to transfer from politically accountable officials to the courts the decision as to which among reasonable alternative law enforcement techniques should be employed to deal with a serious public danger” (id.). Rather, it is sufficient for a court to determine that there is some “empirical data” demonstrating the effectiveness of the means chosen by law enforcement officials (id., at 453-455).
As for the severity of the interference, the Court found that “[t]he intrusion resulting from the brief stop at the sobriety checkpoint [was] for constitutional purposes indistinguishable from the checkpoint stops * * * upheld in Martinez-Fuerte” (Michigan Dept. of State Police v Sitz, supra, at 453). Finally, there was no question in Sitz that the discretion of officers in the field was adequately circumscribed, because the checkpoint was operated according to “guidelines setting forth procedures governing checkpoint operations, site selection, and publicity” (id., at 447).
Respondent in Muhammad F. and defendant in Boswell urge that the absence of a fixed checkpoint or roadblock is alone sufficient to invalidate the seizures in these cases. We do not accept that proposition. The Supreme Court has not adopted a per se rule banning all such stops and requiring a fixed checkpoint or roadblock in all cases. Suspicionless patrol stops *146are suspect as a general matter because of both their elevated potential intrusiveness and their greater opportunities for the unlimited exercise of discretion by police; no such stop has been upheld by the Supreme Court or our Court when it was conducted at random. Suspicionless stops, however, of “all oncoming traffic at roadblock-type stops” to check driver license and registration are permissible (Delaware v Prouse, supra, at 663; see also, id., at 664 [Blackmun, J., concurring; “I necessarily assume that the Court’s reservation also includes other not purely random stops (such as every 10th car to pass a given point)”]; Texas v Brown, 460 US 730, 739).
Consistent with the foregoing approach, People v John BB. (supra, 56 NY2d 482) upheld the constitutionality of patrol stops conducted in a “uniform and nondiscriminatory manner” of all vehicles traveling in a sparsely populated rural area soon after a recent series of burglaries was discovered (id., at 488-489). The “legitimate function of acquiring information regarding the recent burglaries” outweighed the interference with drivers’ rights, given that the area “was too large to permit any meaningful utilization of traditional investigative procedures” (id., at 488). Thus, neither New York nor United States Supreme Court precedents require automatic foreclosure of such stops, without examining whether a systematized non-arbitrary method was employed to accomplish them and conducting a particularized inquiry into the reasonableness of the stops, using the Brown v Texas balancing analysis.
IV.
In applying the arbitrariness and Brown v Texas standards to the circumstances presented on the record before us, we conclude that the stops of the taxicabs in these cases were unreasonable and invalid. Regarding the first factor to be weighed, we do not doubt the magnitude of the governmental interest in protecting victim-prone taxicab drivers late at night on urban streets from a crime wave of violent robberies and homicides by disseminating information to the drivers and preventing crimes that are in progress or imminent.
Turning to the effectiveness prong of the Brown v Texas test, it was, of course, for New York City law enforcement officials to decide “which among reasonable alternative law enforcement techniques should be employed to deal with [the] serious public danger” of violent taxicab crime (Michigan Dept. of State Police v Sitz, supra, at 453 [emphasis supplied]). The prosecution in these cases, however, failed to submit any empirical ev*147idence that this type of patrol stop — as opposed to fixed checkpoint stops by uniformed officers in marked cars — was a reasonably effective means of furthering the State interest in reducing violent crimes against taxi drivers (cf., id., at 454-455; People v Scott, supra, at 523, 526-527). There was no showing of the unavailability of less intrusive or discretionary means to prevent violent crime directed at the drivers of livery vehicles (see, Delaware v Prouse, supra, at 659). Absent from the record is any evidence that the operation of a stationary checkpoint by uniformed officers in marked police cars was “impractical” (United States v Martinez-Fuerte, supra, at 557) or that circumstances did not “permit any meaningful utilization of [such] traditional investigative procedures” (People v John BB., supra, at 488).
Moreover, these stops were excessively and, on these records, unjustifiably intrusive, both objectively and subjectively. As for the objective intrusiveness, the safety checks in Muhammad F. were described as routinely involving a request or direction that the passengers step out of the cab while the officers searched it (cf., United States v Brignoni-Ponce, supra, at 880). Subjectively, in each case, the police employed a roving-patrol stop in which a single, unmarked police car operated by non-uniformed officers pulled over the taxi. Since it was nighttime and there were no signs, cones or flares to warn the drivers that they would be stopped, the motorists were “taken by surprise” (United States v Martinez-Fuerte, supra, at 559). The plainclothes officers then proceeded to walk towards the taxicabs, shining their flashlights at the occupants. Because the cab drivers could not “see * * * other vehicles * * * being stopped” and there were no “visible signs of the officers’ authority,” this is precisely the type of police stop that would “generat[e] concern or even fright on the part of lawful travelers” (Michigan Dept. of State Police v Sitz, supra, at 452-453). Indeed, given that police chose the locations of the stops because they were high taxi-crime areas, the occupants’ first reaction might well have been fear that they were being car jacked or mugged. The subjective intrusion would have been significantly less had there been some means employed to provide assurance that this was a police operation employing “a uniform system for stopping cars” (People v Scott, supra, 63 NY2d, at 527).
Of equal significance, the records in these cases contain no showing that the Task Force had attempted to mitigate the constitutional infirmity of “standardless and unconstrained *148discretion” of “the official in the field” (Delaware v Prouse, supra, at 661) other than the vague and purely conclusory testimony that the officers had verbal instructions to stop taxis “in a set basis and not just arbitrarily.” For example, had the Police Department produced evidence of particularized guidelines with “listed criteria” that “established procedures for site selection, lighting and signs; avoidance of discrimination by stopping all vehicles, or every second, third or fourth vehicle; [and] location of screening areas” (People v Scott, supra, at 522-523), then we would have some assurance that the stops were “being maintained in accordance with a uniform procedure which afforded little discretion to operating personnel” (id., at 526; see, Delaware v Prouse, supra, at 650; cf., Michigan Dept. of State Police v Sitz, supra, at 447, 453; see generally, LaFave, Controlling Discretion by Administrative Regulations: The Use, Misuse, and Nonuse of Police Rules and Policies in Fourth Amendment Adjudication, 89 Mich L Rev 442 [1990]). Since the officers here were not even required to make a written record of stops that had taken place, in conducting our “post-stop judicial review” (United States v Martinez-Fuerte, supra, at 559), we are relegated to the self-verifying evidence from the officers whose conduct is being challenged to determine whether they were using uniform and non-discriminatory procedures.
On the scanty proof adduced in these cases, there was a failure either to establish the reasonableness of the patrol stops here under the three-pronged balancing test of Brown v Texas (supra) or to satisfy the constitutional requirement that the stops were “carried out pursuant to a plan embodying explicit, neutral limitations on the conduct of individual officers” (id., at 51; cf., People v John BB., supra, at 485 [“stops were conducted in a uniform, nonarbitrary and nondiscriminatory manner”]). Thus, the evidence was properly suppressed in Muhammad F. and improperly admitted in Boswell.
With respect to the dissent, the focus of these appeals in our Court was the legality of the stops. The issue of standing was barely mentioned by the prosecutions; in the lengthy Appellate Division decisions it was not alluded to at all. We therefore choose to assume standing for present purposes, consistently with our precedents, and reach the merits of the issues the parties addressed. As for the dissent’s analysis on the merits, we have already noted that the stops in these cases were carried out arbitrarily at the discretion of the officers in the field, whereas in People v John BB. (supra), the stop “was made pur*149suant to a nonarbitrary, nondiscriminatory and uniform procedure, involving the stop of all vehicles” (id., at 488).
Accordingly, in Muhammad F., the order of the Appellate Division should be affirmed, without costs, and the certified question not answered as unnecessary; in Boswell, the order of the Appellate Division should be reversed and the order of Supreme Court, Bronx County, reinstated.