As the Maryland General Assembly prepares to take up the issue of extending the state's controversial automobile pollution tail-pipe testing program, the key questions are how to do it and how often, not whether to do it.

Increasingly, the reliability and cost-effectiveness of the tests are being called into question as auto manufacturers have developed more sophisticated "on-board diagnostic systems."

These sensors and computers are designed to achieve the best mixture of air and fuel and the least pollution.

"We are seeing cars being tested now that are cleaner than before, and the failure rate is dropping," said Bruce Diehl, director of Maryland's vehicle emissions inspection program.

But the tests begun in 1984 to comply with the federal Clean Air Act are likely to continue for the foreseeable future, for fear of losing federal highway funds should they be suspended.

Without action by the legislature, the current program, operated at 10 locations by a subsidiary of British Petroleum under contract to the state, expires next year. The firm supplies facilities, equipment and employes. In return, it receives most of the $9 inspection fee car owners pay.

Task forces set up by the legislature and Gov. William Donald Schaefer have urged the program's continuation.

The task forces recommended, however, that inspections be every other year instead of annually.

The governor's task force also said that cars should be inspected for 20 years instead of 12.

It rejected pleas by the service station owners to decentralize the inspection system by allowing them to conduct the tests.

Political observers say this recommendation is likely to stir the only major auto emissions controversy among the legislators.

"Rather than 'should we?', the argument has been 'how should we?' " said Bob Douglas, Schaefer's spokesman. "The only pressure has been who tests and when."

Diehl said the current system saves the state administrative costs.

He said the tail-pipe tests still catch the dirtiest cars, although fewer are flunking than before.

"There are a lot more high-tech vehicles, and older vehicles are being maintained better," Diehl said in an interview.

By 1995 "or a little later," Diehl said, most cars may have the computerized diagnostic systems, rendering obsolete the current method of inserting a tube hooked up to a monitoring machine into the tail pipe.

Then, he said, hands-on inspections will be only to insure that air-pollution devices aren't tampered with and leaded gas is not used.

Meanwhile, however, Diehl said, so many pre-1981 cars are still on the road without the new technology that the old system should remain in effect for all 1.8 million vehicles registered in the state.

About 12 percent of the cars tested this year flunked inspection, Diehl said. But the rate on cars built in 1981 or later was only about 5 percent, compared with 20 to 25 percent for older vehicles, Diehl said.

He said the state could require less frequent inspections for newer cars, but that would appear to discriminate against less affluent owners of older cars.

The emissions program required areas with significant amounts of ozone pollution to test cars or face loss of federal highway funds, $200 million in Maryland alone in fiscal 1990-92. The U.S. Environmental Protection Agency requires testing programs in 60 urban areas in 32 states, including Maryland, Virginia and the District.

Acknowledging the federal funding threat, neither state task force addressed basic questions of whether the emission testing actually works.

"Many vehicles may be falsely identified as failing the idle test due to inadequate vehicle preconditioning before testing," a trio of General Motors engineers said in a technical paper presented here in May. "The owners of those vehicles are then inconvenienced because they are required to seek unnecessary, and perhaps expensive, repairs."

The EPA continues to defend the test, although some of its own research calls into question its effectiveness. In a study of cars in the Maryland program last year, 38 percent of the "failed" cars passed without any repairs after being warmed up for a longer time.

Larry C. Landman, the EPA official who directed the Maryland study, said the test, while imperfect, "does a reasonably good job of picking out the dirtiest cars. We're constantly looking for better tests."

Added Dick Wilson, director of EPA's office of mobile services, "Obviously, you don't want to fail people who shouldn't fail. There was a problem with late-model cars and how long they should idle before testing. We still think the test is pretty effective."

EPA is conducting additional studies of "preconditioning" of cars and tail-pipe test variability.

"We're moving into a new era of inspection maintenance programs," said Jane Armstrong, senior project manager at EPA's Ann Arbor, Mich., research facility.

She said the program began because car owners were tampering with early antipollution devices that cut down on automobile performance. Now, the devices are harder to tamper with and performance problems have lessened.

"Now, we're seeing a different reason cars are high emitters: sensor failures or a computer problem. The trend is to cleaner cars, but there are very, very high emissions when they do fail."

Roger Johnson, a local crusader against the state tests, suggests the evidence is scanty to support the underlying premise that cars today are the major source of carbon monoxide and hydrocarbon air pollution.

Johnson contends industry is to blame. But Maryland's Diehl replied that auto emissions are "one part of the whole thing. Industry is regulated almost to the limit. There's always controversy on where pollution is from."