Eight months after setting up an innovative computerized system to look for new drugs to combat the AIDS epidemic, a team of researchers from the University of California at San Francisco (UCSF) reported yesterday that they have identified a common prescription drug that blocks a key part of the HIV virus, which causes AIDS.

The drug, the widely prescribed antipsychotic medication Haldol, cannot yet be used for treating AIDS because in its present form it is effective against HIV only in doses far above the safe limits for humans.

But Irwin Kuntz, leader of the UCSF team, called the discovery a critical first step in finding an effective treatment for the disease and said he was optimistic that a drug similar to Haldol or a reformulated version of it could be available within a year for testing against AIDS.

Yesterday's announcement also marks a critical confirmation of the effectiveness of the "rational" drug design strategy pioneered by Kuntz and other chemists, which some scientists have predicted will transform the way that drugs for a wide variety of diseases are discovered and developed.

In the case of Haldol, for example, the computer sytem used by Kuntz enabled his team not only to find a potential AIDS treatment in a class of drugs where no one else had thought of looking but to do it faster than would be possible by using traditional drug-development methods.

"This is a major step forward," said Health and Human Services Secretary Louis W. Sullivan. "In a remarkably quick time this work has located a compound that may lead the way toward useful new drugs for AIDS."

Rational drug design is based on the fact that how a drug works against a particular disease often can be predicted by looking at its molecular structure. A drug is effective when its shape matches that of the disease it is trying to stop. If a drug can fit over a critical component of something like the AIDS viruses, for example, then it can seal off that part of the virus, covering it like a glove over a hand. The more snugly it fits, the more effective it will be.

Traditionally, researchers have used trial and error to discover whether drugs fit certain diseases, screening thousands of compounds in test tubes and animals one by one in a process that can take years and millions of dollars.

Rational drug design, by contrast, takes advantage of new developments in molecular chemistry, which allows scientists to make accurate, three-dimensional images of specific proteins they want to target. Kuntz's team, for example, began with the segment of HIV known as the protease enzyme, which is essential in helping the virus replicate itself. Many studies have shown that crippling the enzyme inactivates the AIDS virus.

The researchers then took three-dimensional computer images of 10,000 promising chemicals and used the computer to see how well they matched up to the image of the protease enzyme. Of all the candidate compounds, Haldol fit into the protease enzyme the best.

"It was just a matter of weeks for us to decide which three or four compounds we were most interested in," Kuntz said. "It is really striking to be able to proceed that quickly from scratch."

Because Haldol is effective against HIV only at doses about 1,000 times higher than are safe for humans, the UCSF scientists are looking at ways of redesigning the molecular shape of the Haldol so that it will fit the protease enzyme even more snugly, allowing it to be effective in much smaller doses.

Kuntz said his group also hopes to began matching its computer library of more than 60,000 chemicals against other parts of the AIDS viruses as soon as three-dimensional images of more segments of the viruses are made.