As the police get tougher with drunk drivers, a whole new legal specialty is cropping up. More and more defense lawyers are devoting their energies to helping those accused of driving drunk to beat the rap.

The reason for the new specialty is clear: A drunk-driving conviction these days is serious enough to warrant spending the $3,500 or so it can cost to try to win an acquittal -- or at least a less serious charge. "Many, many drunk-driving cases are going to trial, because a lot more is at stake," says John Tarantino of Providence, R.I., who writes a monthly column on the subject for the National Association of Criminal Defense Lawyers.

Suspension of a driver's license is almost certain to follow a drunk-driving conviction, and jail terms are mandatory in 16 states, even for a first offense. Iowa requires a $750 fine plus a week in jail; the penalties in Maryland were stiffened as of July 1. In addition, a drunk-driving conviction causes auto-insurance rates to skyrocket. "The social stigma has also increased," notes Donald Nichols of Minneapolis, editor of a best-selling handbook for lawyers on how to defend such cases. "People come in and say they just don't want that in their record."

Defense lawyers try to raise doubts about the impressions of officers on the scene, and often challenge the results of chemical tests of alcohol levels in the blood. Just because a driver's breath smelled of alcohol, for instance, doesn't mean he or she was legally drunk. "It's not against the law to drink and drive," Tarantino reminds clients. "That's not a crime."

Defense lawyers start by questioning whether the policeman's on-the-spot assessment of the driver's condition was reasonable. Usually, that assessment is based on the results of a simple test such as walking a straight line, standing on one leg or picking up coins from the ground. Defense lawyers argue that even someone stone-cold sober might falter on those tests because of some physical infirmity, bad lighting, uneven pavements, or the nervousness engendered by any encounter with the police. Local liquor-store owner Stuart Miller this month won acquittal in a well-publicized drunk-driving trial by contending that he flunked a roadside sobriety test because he wobbled on the high heels of his cowboy boots.

Defense lawyers challenge such assessments for a number of important reasons. For example, if lawyers can convince a judge that the police lacked "probable cause" to arrest a driver, then none of the results of more complex testing done at the station house can be admitted at a trial.

But even if these later tests are admitted, lawyers can use a sheaf of expert reports to challenge the seemingly damning results. Breath sprays can throw off the readings. Liqueurs or similar heavy drinks linger longer in the mouth, giving a high reading on a breath test although the actual level of alcohol in the blood is not at a dangerous level. Alcohol levels tend to rise for an hour or so after the last drink, so the reading taken at the station house may be significantly higher than it was at the time the defendant was stopped. And the machines themselves often malfunction: Some models are thrown off by radio-frequency interference, and others are not maintained in top-notch condition.

One recent case involved a 34-year-old Massachusetts bank vice president, who went to a Red Sox game at Fenway Park although he still suffered the vestiges of a stomach flu. During the evening, he had three beers. On the way home to Rhode Island, a headlight on his car stopped working, and a policeman pulled him over. The tension, beer, flu and medication combined to make the banker vomit; and the suspicious officer took him to the station house, where a breath test indicated a 0.11 percent alcohol level in the blood -- strong evidence of impairment.

Tarantino, however, hired an expert who walked the banker through the same drinking and eating regimen he had followed during his evening at the baseball game, and then, when a like amount of time had passed, administered another breath test. It measured only 0.05 percent -- well below the legal limit. The jury acquitted the banker. The probable reason for the disparity: Vomiting brought alcohol from the stomach to the mouth, producing a breath reading out of all proportion to the actual amount of alcohol in the banker's blood.

But defense lawyers can mount a successful case even when they cannot attack the test results. They often argue that merely because the average person drives poorly with a blood-alcohol level of 0.10 percent doesn't mean that the defendant is impaired at that level. "There's a huge amount of interpersonal variability" in how drinking affects driving, says toxicologist Jonathan D. Cowen of Prospect, Ky., a frequent expert witness in drunk-driving cases. That line of defense is currently barred in a number of states -- including California, Illinois, Pennsylvania and New York -- that call a 0.10 percent blood alcohol concentration absolute proof that the drinker should not have been driving.

Cowen now is involved in a group of cases in Florida that are testing the concept of per se illegality. The six defendants in that merged proceeding contend that preventing them from introducing evidence that might help them at trial violates their constitutional rights. A ruling in their favor would be a major victory for the new legal specialty, and almost certainly would guarantee that a lot more drunk-driving defendants would seek out lawyers to fight their cases.