The Baltimore police call it the "no-good SOB case."

A man was arrested for a shooting. While he was at the station, a Baltimore policeman sauntered in and announced, "I'd like to shake the hand of the guy who shot that no-good SOB."

The suspect stood up, introduced himself, accepted congratulations -- and was later convicted on that evidence.

In Baltimore, they tell the story to illustrate what policemen may or may not do under a recent Supreme Court decision, known as Rhode Island vs. Innis, in which police played on a suspect's conscience to obtain a confession. It may or may not be an accurate illustration, but it's the best they've got.

The problem for the Baltimore police, and for police all around the country, is that as the Supreme Court wrestles with its desire to protect civil liberties and its desire to see that the guilty are punished, its decisions have become harder and harder to interpret.

The issue raised by the Innis case was difficult and important. It could have become one of those famous Supreme Court cases that law students memorize, or that breaks new ground and gives new guidance to police and suspects.

It could have been, but it wasn't. And the reasons it wasn't have less to do with what the court said than how it said it. Three months after the opinion came down, many of the lawyers who need to use it are saying it was incomprehensible, meaningless.

And the lawyers say that Rhode Island vs. Innis is not an exception but rather a tendency. In the field of criminal justice, where police in particular need hard and fast rules, many say they are increasingly getting nuances and gray areas and legal reasoning intelligible perhaps to law professors but often useless in application.

"You have a right to remain silent. If you choose not to remain silent, anything you say or write can and will be used as evidence against you in court. You have a right to consult a lawyer before any questioning and you have a right to have a lawyer with you during questioning."

Every police officer knows these lines -- part of the famous Miranda warning. They also know -- or thought they did until Rhode Island vs. Innis -- the rule that goes with them: Once a suspect says he wants a lawyer before questioning, you're not to question him or her.

On Jan. 17, 1975, in Providence, R.I., police arrested Thomas J. Innis for the murder of a taxicab driver. They gave him his Miranda warning and he did, indeed, say he wanted a lawyer. The three policemen who joined Innis in a paddy wagonn after his arrest were explicitly told by a superior not to question him.

But once inside the wagon, the officers began to talk about the murder weapon -- a shotgun -- which was still missing. It could be a critical piece of evidence.

The officers talked among themselves, not questioning Innis at all, but conversing loud enough for him to hear.

"I frequent this area while on patrol," one officer said. "There's a lot of handicapped children running around in this area and God forbid one of them might find a weapon with shells and they might hurt themselves."

"It would be too bad if a little girl would pick up the gun, maybe kill herself," said another officer.

Apparently conscience-stricken, Innis told the officers to turn back; he would lead them to the weapon. He did and it was used to convict him in court.

The justices of the Supreme Court knew the questions to be answered in Rhode Island vs. Innis. Of course, interrogation is improper when a suspect who wants a lawyer doesn't have one. But the conversation among police in the paddy wagon was not clearly an interrogation. Could a conversation among police be considered an interrogation?

The justices also knew the importance of their answers. They had seen cases like this before: sophisticated police officers -- with wit, guile or serendipity -- gaining information from an uncounseled suspect.

On May 12, 1980, the court ruled. An interrogation, Justice Potter Stewart wrote for a five-man majority, does not have to be "direct" questioning. It "refers not only to express questioning, but also to any words or actions on the part of the police . . . that the police should know are reasonably likely to elicit an incriminating response from the suspect."

The definition focuses on the "perceptions of the suspect, rather than the intent of the police," the court said.

It sounded like Innis had won. But when the court turned to the specifics of the case the confusion began. The justices concluded that "the entire conversation" among the police transporting Innis "consisted of no more than a few offhand remarks."

It appeared that Innis lost. His case was sent back for further proceedings.

In Lincoln, Neb., Gerald F. Fisher, legal adviser to the city police department, began preparing a memo on the Innis case to keep his department informed.

"I came to the conclusion I really didn't know what to tell them. The court reasoned up to a point and then they reached the wrong conclusion about Innis.

"It was a model of confusion. I really don't know what they did."

In Baltimore, state public defender Alan Murrell began studying the case for the benefit of his staff. " couldn't do it," he said. "It wasn't intelligible. The whole thing was doubletalk."

And last week, also in Baltimore, police agent Robert Oros attempted to teach 30 or 40 city police officers about Rhode Island vs. Innis. But he didn't understand it himself, though he is in law school.

"What matters is whether the suspect thinks it's a question," Oros said. "They're saying 'put yourself in the suspect's position,' But that's almost impossible to do." The real problem, he said, is that after formulating a definition that seemed favorable to Innis, the court turned around and went against him.

When Oros presented Rhode Island vs. Innis to his class of officers last week, he stuck closely to the opinion's wording, reading it from a summary.

Afterward, one of the students, patrolman John Sheehan, was interviewed about the case. Would it be all right to knowingly cook up a story in front of a suspect specifically designed to elicit a certain incriminating response?

"Sure it would be okay," Sheehan replied. "The decision didn't change a thing. A smart burglar, a hit man, he's not going to be tricked anyway, but we're allowed to try it."

But what about the court's definition of "interrogation?" Sheehan was asked. What matters, he said, is that the police in the Innis case were upheld by the justices, who, it seemed, didn't think they had used deliberate trickery.

Did they use trickery? Sheehan was asked. "Sure," he said, laughing. "They [the justices] didn't know it, but you and I know it."

Lawyers and legal advisers who tell police how to behave say they look for hard rules, something that can be boiled down into a phrase and drummed into the heads of sometimes reluctant students -- police officers who even then don't always follow them.

In the days when the Warren court was making breakthroughs in criminal justice, it was relatively easy. The rules were clear: Read him his rights. jAllow him a lawyer. Get a warrant for most searches.

Since then, the high court has been modifying and altering, rarely revolutionizing, and things have become a lot trickier.

Consider, for example, warrantless searches involving cars.

You can search a car stopped in connection with a suspected violation. You may search the glove compartment if it's unlocked but you can't unlock it. If the trunk is open, you may search that. If you find a locked box in the trunk, you may not open that. If it's open, you may be able to search it. If a box is locked and somehow you search it anyway, the evidence might be admissible in court if it relates to the passengers of the car. It might not be admissible if it relates to the operator or owner.

Here, for example, is an exchange in Oros' class for Baltimore police officers last week:

Oros: "Suppose you have a suitcase. You make an arrest. Can you take it down and open it?"

Officer: "No, you can't open it."

Oros: "Suppose the guy's holding it. Can you search it then?"

Officer No. 1: "Sure. It's a 'Chimel' search" (referring to the name of a Supreme Court case).

Officer No. 2: "No. You deal with it the same way you deal with a purse. You take it away and then you don't have to search it."

Oros: "I don't know what the answer is. It's a gray area. The courts have said that once you remove it you can't search it. But it's going to depend on what you're arresting them for, if you think there's a gun in it."

The uncertainty is not limited to police. Defense lawyers interviewed are having the same problems and blame them, in part, on confused and incomplete Supreme Court decisions.

Albert Krieger of Miami, president of the National Association of Criminal Defense Lawyers, said that increasingly legal points can be argued with "equal persuasiveness" by both sides using the same Supreme Court opinion.

"It shouldn't be that way," he said. "We're entitled to better guidance than that."

Gerald Gold, a prominent Cleveland-based criminal defense lawyer, said that in some areas of the law "you used to feel that you may not win at trial level but the court of appeals will take a look. Now we say the Supreme Court isn't so clear on this issue. They've been jumping back and forth and we're not sure how to handle it."

The ability of the court still to generate major change in the judicial system with clear opinions is unchallenged.This year and last, for example, the court has been clearly saying that evidence seized illegally from one person may be used to convict another.

For instance, the court upheld the conviction of a drug offender despite the fact that the drugs used as evidence were taken improperly from someone else's purse, where they had been stashed.

The result of this line of rulings, defense lawyers say, has meant more victories for prosecutors in drug and gambling conspiracy cases. And while the defense lawyers don't like it, they say that at least they understand it.

Other areas are gray, criminal lawyers say, leaving lower court judges free to play out their own choices.

The most critical of these areas concerns the use of the controversial "exclusionary rule," which allows judges to throw out evidence taken in an illegal search.

While many expected the Burger court to do away with the exclusionary rule, it has only narrowed it, chipped away at it and, on a regular basis, expressed hostility to it as an impediment to finding the truth in criminal cases.

Defense lawyers say the tone of these rulings is having as much impact as their substance, which rarely goes as far.

"The tenor becomes as important as the words" of the court, said Gold. "It took a number of years for the decisions of the Warren court to have an effect at the trial level. Now, the tenor is that we aren't going to do those things anymore, even though there may have been no ruling saying just that.

"The trial judges say these guys [the justices] really mean there shouldn't be an exclusionary rule."

"The tone has a tremendous psychological impact in lower courts," said Washington lawyer Sol Rosen, who handles every type of crime, high and low, for every type of defendant.

"I really feel the impact," Rosen said. "If you get a hostile judge, one who's hostile to defendants anyway, all he has to do is cite case A or case B and he rules. You talk about sophisticated points of law and they don't know what the hell you're talking about."

"Police are aware of the disfavor in which the exclusionary rule is held at the courts," said Krieger, "and they're pushing it. They're more willing to take chances than they used to be."