The Department of Veterans Affairs said it was investigating the deaths of patients in research projects at hospitals in Detroit, Albany and Fargo, N.D. The department also said it had found "serious noncompliance" with federal rules at its hospitals in Pittsburgh; Providence, R.I.; Martinez, Calif.; and Long Beach, Calif, and detected problems at hospitals in Northampton, Mass., and Portland, Ore.

The new director of the veterans research program, Dr. Nelda P. Wray, ordered the review last month after learning of cases in which researchers falsified data and did not tell patients about the risks of experimental treatments. "Those practices will not be tolerated," Dr. Wray said.

Research is one of the principal missions of the Department of Veterans Affairs, which spends nearly $1 billion a year on about 15,000 studies involving 150,000 patients. Ethics committees are supposed to monitor each study to protect patients.

An internal investigation by the department found that a huge overdose of a prescription drug "led to the death of a research participant in a clinical trial" at the veterans' hospital in Detroit last year.

The department has joined federal prosecutors in a criminal investigation of two researchers who worked at the veterans' hospital in Albany. Fabrication of data in those studies may have "contributed to the death of one or more patients," a memorandum from the department says.

At the veterans' hospital in North Dakota, all clinical research was halted on Feb. 28. Federal experts who inspected the hospital said one patient had died and 22 had suffered adverse reactions to drugs in research projects since September. The adverse reactions were not promptly reviewed by the hospital, according to the veterans agency's office of research compliance and assurance, a watchdog body.

An official report on the Fargo hospital says that a doctor there was listed, "without his knowledge or consent," as being responsible for patients in a clinical trial comparing drugs for high blood pressure. In a letter suspending research there, Dr. Jonathan B. Perlin, deputy under secretary of veterans affairs, said the study, involving 87 patients, was not properly supervised.

Some hospital administrators said they did not have enough money or employees to comply with all the federal requirements.

Doctors at veterans' hospitals have done pioneering research on spinal cord injuries and schizophrenia, helped develop the implantable cardiac pacemaker and the nicotine patch, and performed some of the first successful liver transplants. The agency has also been a leader in analyzing medical errors.

But officials at those hospital now say they are reluctant to discuss possible improprieties because they fear being sued. Jayne M. Steubing, the widow of the Albany patient, Carl Steubing, sued last month, asserting that researchers had falsified test results and improperly enrolled him in the trial of a cancer drug, hastening his death from stomach cancer.

"Our whole purpose is to ensure that this kind of thing does not happen to other people," Mrs. Steubing said in an interview.

Investigators said the Detroit patient, Cyril V. Krcmarik, died after a series of medical errors.

The Department of Veterans Affairs and the Food and Drug Administration faulted researchers who treated Mr. Krcmarik for prostate cancer. He received a large overdose of dexamethasone, a steroid to combat side effects of chemotherapy.

Mr. Krcmarik lived near Saginaw, Mich., so he was allowed to fill his prescription at the veterans' hospital there. But, investigators said in a recent report, the prescription was "incorrectly rewritten" by a doctor at the Saginaw hospital and the pharmacist further changed the instructions, and as a result the patient used a nine-month supply of the drug in less than three weeks. Moreover, the report said, the government twice refilled Mr. Krcmarik's prescription and thus "missed opportunities" to correct his dosing schedule before he died on April 2, 2002.

Mr. Krcmarik's daughter-in-law, Claudia S. Krcmarik, said in an interview: "The veterans' hospital appeared to be a research machine, using these elderly patients without enough oversight. This is a vulnerable age group. You have people who are hard of hearing, may have difficulty understanding instructions and tend to trust their doctors."

Ann Talbot, a spokeswoman for the Detroit hospital, said: "Somebody died. That was awful, and we are very seriously concerned."

The researcher directing the study, Dr. Joseph A. Fontana, said: "The patient was given more drug than he was supposed to get. That was a pharmacy error. It will never happen again. I am doing things differently now."

Federal officials told researchers at the V.A. Northern California Health Care System, in Martinez, that they could not recruit or enroll any new patients because they had not adequately protected veterans in clinical research.

Inspectors from the Department of Veterans Affairs said the California researchers did not properly assess the safety of experimental treatments, overstated the benefits and did not adequately explain the risks to patients, many with Alzheimer's disease and other brain disorders. P> Dr. Kenneth W. Kizer, who was under secretary of veterans affairs from 1994 to 1999, said: "The problems at the V.A. are not unique. Some prestigious universities have similar problems. Clinicians view a lot of the federal requirements as administrative hassles and do not take them seriously."

But the chief of the agency's research compliance office, Dr. John H. Mather, said: "The rules are important. They reflect the ethical underpinnings for the conduct of research, the need to treat other human beings with dignity and respect."

The agency's Dr. Wray said all employees engaged in clinical research would have to take courses in research ethics and "good clinical practice." In addition, she said, all researchers must have their credentials verified once a year.

Federal officials said such verification procedures might have detected problems with one of the Albany researchers, whose medical license was revoked by Iowa and Pennsylvania in the early 1990's.