The Nuclear Regulatory Commission yesterday issued a new report on the potential consequences of accidents at atomic power plants, but did not release the worst-case death and damage estimates made by the study, citing the "very low probability" of such events.

The Washington Post reported yesterday that the study, performed by the Department of Energy's Sandia National Laboratories, had concluded that the worst-case death toll from an accident at certain atomic power plants could exceed 100,000 and damage could top $300 billion.

Robert Bernero, director of the NRC's division of risk analysis, told a news conference yesterday in Bethesda that the worst-case calculations "made in conjunction with the study" were not included in the report because "they represent consequences of accidents whose probabilities are extremely low."

"The combination of this unlikely accident, together with combinations of very unlikely weather conditions, can lead to calculated consequences having probabilities of about one in 1 billion per year of reactor operation," Bernero said. "The results in the report released yesterday by the NRC did not present consequences whose probabilities were lower than 1 in 100 million."

The Union of Concerned Scientists, however, at a news conference in Boston, sharply criticized the NRC for withholding the results of the study for more than six months and then issuing the report "without including the peak estimates."

"We certainly agree these are worst-case estimates, and the accidents are very low probability events that we hope will never happen," said Eric Van Loon, executive director of the Cambridge-based scientific group. "But I think it's worth remembering that before Three Mile Island, the accident that occurred there was officially termed 'not a credible event.'

"It seems to me the NRC is once again feeding selective data to the public on the theory that they know best what information the public should have," Van Loon said. "I think people would have more confidence in our federal nuclear regulators if they did not repeatedly try to dismiss or minimize the dangers of nuclear power, while overwhelmingly professing the benefits."

The Atomic Industrial Forum, a trade association that represents the nuclear power industry, said yesterday it did not have any comment on the new NRC study.

The Nuclear Regulatory Commission, at its press conference, stressed the statistical improbability of a "Group 1" accident -- one in which a core meltdown leads to a large release of radioactivity into the atmosphere -- occurring at a time when winds would blow it over a nearby city, where rain would then make it "fall on people's heads."

Bernero said he felt the chances of this happening were less than the possibility of a jumbo jet crashing into a football stadium during the Superbowl.

But Rep. Edward J. Markey (D-Mass.), chairman of a House subcommittee that oversees the NRC, yesterday was sharply critical of Bernero's use of this analogy. He said it was "unfortunate that the NRC has reverted to the role of being the protector and promoter of the nuclear industry by restating such misleading examples."

"The Sandia report is primarily a study not of accident probabilities but of possible accident consequences," Markey said. "The results prove beyond a shadow of doubt that the worst-case consequences could be far in excess of what anybody ever thought they would be."