Government officials ignored warnings from experts that a major nuclear safety study had serious flaws and kept the project going for years to counter potential compensation claims from radiation-damaged workers, according to Energy Department documents.

However, in the mid-70s, when results of the study unexpectedly showed danger from low-level radiation, disappointed Atomic Energy Commission officials yanked the funding for the research before it was made public and decided to shift the study to in-house investigators critical of the findings.

The Energy Department documents were turned over late last year under a Freedom of Information request by Dr. Thomas Mancuso, a University of Pittsburgh epidemiologist who originally headed the government study, and Michael Bancroft, an attorney for the Public Citizen Litigation Group here. The documents were made available to The Washington Post.

Low-level radiation has become an increasing concern among some medical experts who fear it may be responsible for elevated cancer levels. Nearly 500,000 past and present employes at private and government unclear installations have been exposed to such radiation and hundreds of thousands of persons received radiation exposure during government nuclear tests in the southwest since the 1940s.

A federal researcher said yesterday that the number of workers exposed to the radiation could be 5 million.

Mancuso, a leading radiation epidemiologist, was granted federal funds in 1964 for what turned out to be the government's major long-term study of low-level radiation. The $6 million project was aimed at radiation exposure in workers at the government's nuclear facilities at Hanford, Wash., and Oak Ridge, Tenn.

The Energy Department documents give the following information about the study:

In 1967 a panel of radiation experts picket by the AEC warned that the study's objectives were "hopeless" because too few workers were included. But several members of the group pointed out that since the "political" purpose of the work was to fend off workers' claims against the AEC for radiation damages, the study should be continued.

When a separate study in 1974 by Dr. Samuel Milham, a Washington state epidemiologist, did show elevated cancer levels in the Hanford workers, AEC officials tried to pressure Mancuso into contradiciting the Milham study. The AEC effort was made despite recommendations from its own consultants two years earlier that the real value from Mancuso's study lay in its still-incomplete long-term conclusions.

A year later federal nuclear officials were stunned to learn that Mancuso also had discovered long-range radiation porblems in the Hanford workforce. The federal agency terminated Mancuso's research contract when he refused to contradict Milham's study.

Senior officials in charge of the federal study ordered a "clandestine" search for a replacement for Mancuso to avoid possible government embarrssment. The work eventually was turned over to a group of researchers who had been sharply critical of Mancuso's study.

According to the documents, the 1967 AEC review panel was unanimous in its rejection of the scientific worth of the Mancuso study. The study "does not have, never did have, and never (in any practicl sense) will have any possibility of contributing knowledge of radiation effects in man," wroter one reviewer.

Still, the AEC decided to continue the project. "The study probably will not confirm or refute any important hypothesis but shoud permit a statement to the effect that a careful study of workers in the industry has disclosed no harmful effects of radiation (if the results are negative, as they are likely to be," wrote Sidney Marks, the AEC contract officer for the study.

"That statment, supported by apporpriate documentation, would seem to justify the existence of the study," Marks added.

In 1976 Marks left the government and went to Battelle Laboratories, where he was appointed by officials of the Energy Research and Development Administration, the AEC's successor, to take over Mancuso's work.

"What happended," Mancuso said in a telephone interview yesterday, "was that the AEC set out to fund a political study with guaranteed negative findings. When they foundout their political purpose had collapsed, they dumped me."

After he was dropped from the study, Mancuso released findings in 1976 showing a 6 percent increase in certain cancers among the Hanford workers.

DOE officials first criticized the Mancuso report as the "result of an inappropriate use of statistical methodology" but later acknowledged it raised serious concern about the adequacy of the government's radiation exposure standards.

Last May, President Carter ordered Health, Education and Welfare Secretary, Joseph A. Califano Jr. to lead an interagency group to formulate a coordinated radiation program. Environemental groups and unions representing radiation-exposed workers have sought to remove DOE from the study because they said its role in the Mancuso affair showed a nuclear bias.

Gene Moss, a health researcher with the National institute for Occupationa Safety and Health, said the group had turned up evidence of leukemia and other cancer in workers and others exposed to radiation at levels well below federal safety standards.