The federal government is trying to lure pharmaceutical companies into investing in riskier medical research by offering to pay for and carry out early clinical trials of experimental drugs.

The clinical trials are part of an unusual effort by the National Institutes of Health to solve a seemingly intractable problem: the dearth of breakthrough drugs for diseases that have long stymied researchers. The shortage stems from drug companies' aversion to investing in promising but untested ideas that have been cooked up in academic labs. The cost of getting a new drug to market has ballooned, and drug makers tend to pursue products where the biology is known and the pathway to regulatory approval is laid out.

The project could bring pure research out of university labs and into the marketplace, but it faces hurdles in drug makers' reluctance to share information with potential rivals.

Novel drugs present other problems for drug companies. They aren't sure how to structure their clinical trials to prove to regulators such as the Food and Drug Administration that their drugs work. Basic science, often funded by the NIH and carried out at top universities, rarely addresses these needs, leaving what's called a "translational gap" that stalls progress toward new medicines.

"The hardest part about getting a totally new drug for a new indication (symptom or medical problem) to market is that no consensus exists over how to prove it works," said Wayne Fenton, who leads the NIH project. "The FDA doesn't want to just accept what it's told by the drug companies. As a federal agency, we are in a unique position to broker a scientific consensus."

The effort shows the NIH expanding its role from cutting checks to universities for basic research to more broadly identifying and fixing structural problems in U.S. medical research that impede advancements in science.

The program is starting with schizophrenia, a debilitating mental illness that hasn't had a truly new drug-treatment approach in 50 years. The NIH project invited drug makers to participate in a clinical-trial network that would clear away some of the cost and risk of drug development for the companies, and would give them access to top government and academic scientists. In all, 24 drugs were offered, mostly by small companies, but drug giant Eli Lilly & Co. also raised its hand. Two compounds to treat schizophrenia, developed by Organon Inc. and Targacept Inc., were picked to be tested later this summer.

The federal government has stepped in to support mostly later-stage clinical trials for drugs already on the market in areas of great need, like cancer and AIDS. But it's rare for drug makers to hand over early-stage drugs, as they are in this program.

The project's cost for the NIH, at $13 million, is small for the agency, which has an annual budget of $28 billion, and for the drug industry. But if the project succeeds, major drug makers probably would pour investment into the new trials' target area, schizophrenia and cognitive problems, because of the huge market potential. The drugs currently used to treat schizophrenia, known as antipsychotics, are the industry's fourth-biggest class of drugs, with world-wide sales of about $12 billion last year. Public-private partnerships like this one could bring hope for new treatments in other diseases that have frustrated researchers, including depression and Alzheimer's disease.

But the NIH's foray into drug development has met with skepticism from Pfizer Inc. and other drug makers who prefer to keep control of their research so as not to tip off competitors about what they are working on. Even with the companies that have donated their experimental drugs, the leaders of the NIH project have had to overcome concerns about protecting proprietary and competitive interests.

After long negotiations, Organon and Targacept agreed that if one of the drugs works under the new program, the company will take it forward and retain commercial rights. In addition, the results will be made public, so other scientists and companies can learn from them and use the concepts in their own research.

The need for new therapies in schizophrenia is immense. Antipsychotics ease the delusions and paranoia that patients experience, but don't alleviate the less-known cognitive symptoms, which rob patients of their ability to concentrate and remember and process information. But it is these symptoms that keep patients crippled; only one in five people with schizophrenia is able to live independently or return to work or school.

Frank Baron hasn't been able to resume his old life since he developed schizophrenia. At age 24, he was a Peace Corps volunteer in Yemen helping government officials write software for engineering projects. At age 44, Mr. Baron can't write a line of computer code because he can't concentrate or solve problems. And although he renews his registration as a licensed civil engineer in California every year, he hasn't worked in that field for more than a decade. "I don't have the heart to let it go," Mr. Baron says.

The first part of the NIH project got under way in April 2003 when neuroscientists, psychiatrists and drug-company employees gathered to debate and set priorities for research. Executives from major companies like Lilly and Merck & Co. spoke frankly about the obstacles to drug development. Regulators from the FDA answered questions about what companies would have to do to get approval for drugs to treat the cognitive aspects of schizophrenia.

Next the NIH tackled one of the biggest obstacles to drug development: choosing which of the more than 100 tests now available should be used in clinical trials to measure memory, attention and other brain functions. At the time, individual drug companies used wildly divergent tests, and competing university researchers often invented and patented their own testing methods, and licensed them for sale.

The hard-core scientists wanted an exhaustive battery with newer tests, while drug-company officials said it wasn't feasible to make people in clinical trials complete hours of tests. In the end, the panel whittled it down to a 63-minute battery of 36 tests for use in clinical trials.

With that hurdle out of the way, Stephen Marder, a psychiatrist at the University of California at Los Angeles who won the NIH grant to run the project, set out to convince companies to submit their drugs to the clinical-trial network. He placed advertisements in medical journals that included assurances that the companies would retain all the patents and intellectual property on the drugs.

To apply, the drug companies turned over information that wouldn't usually be let outside company walls, including early animal and lab tests, toxicology information, and descriptions of how the drug worked.

Pfizer, which puts about 20 percent of its $7.7 billion R&D budget into diseases of the central nervous system, stayed on the sidelines. "We can do the work ourselves so we are not seeking outside help," says Atul Pande, vice president of Pfizer's Neurosciences Development. "We have reservations about the intellectual-property and competitive issues."

Targacept, a tiny private company in North Carolina backed by venture capitalists, was working on drugs to activate the brain's nicotinic receptors, which are thought to play an important role in cognition. The company had a trove of research passed on when it was spun off from R.J. Reynolds Tobacco Co. in 2000. Many people with schizophrenia are inveterate smokers, leading some to speculate that the habit was a form a self-medication.

But Targacept was already testing its drug in Alzheimer's disease and couldn't afford to do schizophrenia trials. Chief Executive Donald deBethizy estimated that an initial trial of the drug in schizophrenia would cost $5 million. He would love to have the government pick up some of the tab, but was initially wary.

"Are we going to have enough influence over the study design to be comfortable with the outcome?" Dr. deBethizy asked. He also worried that participating in the NIH project would damage Targacept's chances of getting a big pharmaceutical partner to help it commercialize its drugs. Giant drug makers might see the project as an unpleasant entanglement that reduced the value of the asset, said Dr. deBethizy.

Dr. deBethizy eventually decided to apply. "This research program would sit on the shelf otherwise," he said.

Organon, the New Jersey-based unit of the Dutch conglomerate Akzo Nobel NV, had already done animal studies and a small clinical trial in humans of its drug, known as an ampakine. The drug increased levels of the brain chemical glutamate.

As Organon and scientists planned the trial, seemingly basic details turned into struggles for control. Where would the data be compiled and stored? The companies wanted to maintain the database themselves to ensure that the data met rigorous FDA standards. The investigators said they were perfectly capable of running the database at UCLA.

"From a company perspective, that is not attractive," said David Nicholson, executive vice president of research at Organon.

They also haggled over when the company would get access to the trial results. Organon wanted access within days of the trial's end because, if the drug actually worked, each day of delay would represent lost revenues. Dr. Marder wanted to analyze and finalize the results before turning them over, to preserve the independence of the study.

After months of negotiations, the two sides agreed that the company would get quick access to the data, but they are still negotiating over the database. Organon's Dr. Nicholson doesn't see any of the issues as insurmountable: "This thing is going ahead. We all want to make this work."