MORE THAN 23,000 people were killed in alcohol-related traffic accidents in 1988. While that figure is depressing enough, it's a very large improvement over the figures of only a decade ago. A number of factors have contributed to a reduction in the rate of drunk driving, among them changing attitudes toward alcohol in general, citizen pressure for increased law enforcement, and education -- particularly efforts aimed at young drivers. Police officials say that the institution of sobriety checkpoints has also made a difference. Not only are drink-impaired drivers caught at these roadblocks, but it is assumed that many others are deterred from driving by the threat of detection. On Thursday, the Supreme Court found that sobriety checkpoints are effective and pass constitutionalmuster.

As a general rule, police may not stop a car without at least a reasonable suspicion that individuals in the vehicle may be breaking the law. This rules out random stops to search for drivers who have been drinking, as well as stops based on discrimination, abuse of authority or an officer's generalized belief that something may be amiss. Different rules, though, apply when individuals are not singled out, when everyone is stopped and questioned as people are, for example, at truck weigh-in stations or immigration control stations. In these cases, courts look at three factors in deciding whether the government's conduct is reasonable: the importance of the state's interest in establishing the checkpoint, the effectiveness of the procedure and the level of intrusion on individuals' privacy rights. There is no argument over the magnitude of the drunk driving problem or the state's interest in addressing it. Three justices questioned the effectiveness of checkpoints and the invasion of individuals' privacy. But Chief Justice Rehnquist, writing for the majority, emphasized that the rate of alcohol-impaired drivers found -- 1.5 percent of those stopped in Michigan, where the case originated -- was significant, while the inconvenience to the public -- stops averaged less than 30 seconds -- was slight.

When checkpoints were set up in Maryland eight years ago, police frequently found that sober spouses or friends were driving drunk passengers who might otherwise have been at the wheel. Taxi companies reported a substantial increase in business from intoxicated riders. This is anecdotal evidence, of course, that checkpoints are effective, but there are no statistics on drinkers who decide not to drive. The overall numbers on the decline of alcohol-related fatalities is nevertheless persuasive. Because sobriety checkpoints are nondiscriminatory and minimally intrusive, they are a reasonable response to behavior that is lethal.