A landmark government-financed study that compared drugs used to treat schizophrenia has confirmed what many psychiatrists long suspected: newer drugs that are highly promoted and widely prescribed offer few - if any - benefits over older medicines that sell for a fraction of the cost.

The study, which looked at four new-generation drugs, called atypical antipsychotics, and one older drug, found that all five blunted the symptoms of schizophrenia, a disabling disorder that affects three million Americans. But almost three-quarters of the patients who participated stopped taking the drugs they were on because of discomfort or specific side effects.

One of the newer drugs, Zyprexa, from Eli Lilly, helped more patients control symptoms for significantly longer than the other drugs. But Zyprexa also had a higher risk of serious side effects - like weight gain - that increase the risk of diabetes.

The study, released yesterday and to be published Thursday in The New England Journal of Medicine, was widely anticipated because it is by far the largest, most rigorous head-to-head trial of the newer antipsychotics conducted without significant drug industry financing. The new drugs account for $10 billion in annual sales and 90 percent of the national market for antipsychotics.

The findings may not significantly alter the prescribing patterns of doctors in private practice, who often do not have to worry about cost, psychiatrists said. But they are likely to have an enormous effect on state Medicaid programs, many short on funds in part because of the high cost of schizophrenia drugs.

Several states, including Kentucky, West Virginia and Maine, have limited access to newer drugs, which cost 3 times to 10 times more than the older drugs. "The new study presents an opportunity but also a risk," said John Goodman, president of the National Center for Policy Analysis, a policy research group based in Dallas, which estimates that Medicaid programs spend at least $3 billion a year on antipsychotics, more than for any other drug class.

"The opportunity is to lower the cost of these drugs," Dr. Goodman said. "The risk is that state Medicaid programs use this excuse to entirely deny some patients access to more effective and more expensive drugs which work for those patients."

The government study set out to judge each drug by how long patients and their doctors continued the therapy, a criterion rarely used in studies by drug makers but crucial in real-world practice. People with schizophrenia struggle with delusional thoughts, private voices, blunted emotions and other symptoms, and most try multiple drugs in trying to avoid severe side effects.

The researchers, led by psychiatrist Dr. Jeffrey Lieberman, then at the University of North Carolina and now at Columbia University, recruited 1,493 people with the disorder and assigned them to receive one of five drugs: Risperdal, from Johnson & Johnson; Seroquel from AstraZeneca; Geodon from Pfizer; Zyprexa; and an older drug,perphenazine.

After 18 months, the researchers found, 64 percent of the patients taking Zyprexa had stopped, and at least 74 percent had quit each of the other medications. The most common reasons were that the drug was not effective, the patient could not tolerate taking it, side effects like sleepiness and weight gain or neurological symptoms like stiffness or tremors.

Doctors' concerns about neurological side effects in particular have sped the switch to newer schizophrenia drugs over the last decade. Studies have shown that these medications carry a lower risk than the older drugs of tardive dyskinesia, a disorder that causes tics, lip-smacking and other involuntary movements.

But the study found that at more modest doses, the older drug, perphenazine, while just as effective, was not significantly more likely to cause neurological symptoms. Dr. Lieberman said that there was no reason to believe that modest doses of other older drugs, like Haldol, would perform differently.

The patients on Zyprexa were less likely to be hospitalized because their condition worsened than those taking the other drugs, the study found. But these patients also gained the most weight, adding an average of two pounds a month while on the drug, and their lipid levels increased more than those of people on the other drugs. Weight gain and elevated lipids are risk factors for diabetes.

In the doses used in the study, a month's supply of perphenazine costs about $60, compared with $520 for Zyprexa, $450 for Seroquel, $250 for Risperdal and $290 for Geodon, according to Drugstore.com.

"Probably the biggest surprise of all was that the older medication produced about as good an effect as the newer medications, three of them anyway, and did not produce neurological side effects at greater rates than any of the other drugs," said Dr. Lieberman in an interview.

Dr. Robert Baker, who directs the neuroscience group at Eli Lilly, said that he was pleased with the findings. He said the weight gain and other side effects of Zyprexa were "very consistent with what we've seen in our studies" and that the company tells doctors about these symptoms.

"I think what we can conclude from this study is that there is no one-size-fits-all treatment for schizophrenia," Dr. Baker said.

Spokesmen for Johnson & Johnson and AstraZeneca said that the study supported their drugs and the importance of tailoring treatment to individual patients. "The efficacy results for Risperdal did not demonstrate the full efficacy of Risperdal because many patients in the trial received doses that were too low," Dr. Ramy Mahmoud, of Janssen Pharmaceutica, the unit of Johnson & Johnson that makes Risperdal, said in an e-mail message.

One thing that all agreed on was that the current state of schizophrenia treatment leaves a lot to be desired, and that the field longs for new and different drugs.

"The message is the glass is half full," Dr. Lieberman said. "The drugs work but they are not satisfactory to many patients, and three-quarters of the people in our study voted with their feet and discontinued the drugs."