Late last fall, the Census Bureau assigned a 29-year-old demographer to update the government's population estimate for Iraq. Beth Osborne Daponte quickly found herself drawn into one of the most sensitive political questions of the Persian Gulf War.

How many Iraqis died during the war and its aftermath? The answer, officially taboo in the Bush administration, was indispensable to Daponte's calculations. In January, when a reporter asked for her estimates, she told him: 86,194 men, 39,612 women and 32,195 children died at the hands of the American-led coalition forces, during the domestic rebellions that followed, and from postwar deprivation.

Wednesday evening, after weeks of turmoil during which she was removed from the Iraqi project and her files disappeared from her desk, Daponte was told she is to be fired. Barbara Boyle Torrey, her boss at the bureau's Center for International Research, wrote that Daponte's report included "false information" and demonstrated "untrustworthiness or unreliability." She also accused Daponte of refusing to cooperate with a security clearance investigation.

The White House and Pentagon consistently have sought to suppress discussion of Iraqi casualties, directing analysts and military officers not to provide estimates or professional judgments. Defense Secretary Richard B. Cheney said the day the war ended that "we have no way of knowing precisely how many casualties occurred" during the fighting itself, and predicted that "we may never know."

"I think that Beth is collateral damage in the government's campaign to avoid discussing the question of Iraqi casualties," said William M. Arkin, a former intelligence officer who now does military analysis for Greenpeace. "I think this is an ugly case of retribution."

Frank Hobbs, Daponte's immediate superior, declined to comment. Karen Wheeless, a bureau spokeswoman, said retribution "was not the reason" for Daponte's removal but that she could not discuss the case without violating Daponte's privacy. "Any of us, when we're in trouble, we don't want to look at ourselves as the reason for our trouble," Wheeless said. "That's just human nature."

Daponte, a GS-11 employee, said yesterday she is seeking advice from the American Civil Liberties Union and private lawyers. An aide to Rep. Tom Sawyer (D-Ohio), who chairs a House subcommittee that oversees the census, said Sawyer had been following Daponte's case and planned to investigate her firing.

"Certainly if what she is alleging is true, it would be enormously disturbing," the aide said.

Daponte's firing is based officially almost entirely on a disagreement between the demographer and retired Army colonel Trevor N. Dupuy, a military historian she interviewed as part of her research. Dupuy told census officials, and confirmed in an interview yesterday, that he did not agree with assumptions about civilian deaths that Daponte attributed to him.

Daponte, who showed her handwritten notes of the conversation to a reporter, said that if she misrepresented Dupuy it was an honest mistake. Dupuy said in the interview that he had "no basis" to believe Daponte had "deliberately distorted what I said."

But in the notice of termination, Torrey described Daponte's reliance on "false information" as "a major violation of trust, for which removal is the only effective sanction."

Experts in federal employment law said it was highly unusual for the government to fire an employee in these circumstances without trying to resolve the discrepancy between Daponte and Dupuy. "She's not the first federal employee to make a mistake, if that's what occurred, and not everybody who makes a mistake gets fired," said Joseph Sellers of the Washington Lawyers Committee for Civil Rights. "It smacks of either retaliation or a whistleblower type of phenomenon, where a person discloses something an agency would prefer not to be aired publicly."

Daponte had no access to classified information in preparing her study. She based it instead on a review of literature on casualty modeling and on the gulf war. Her estimates -- a total of 158,000 Iraqi dead, including 40,000 direct military deaths, 13,000 immediate civilian deaths, 35,000 postwar deaths in the Shiite and Kurdish rebellions, and 70,000 deaths due to the public health consequences of wartime damage to electricity and sewage treatment plants -- fall generally within the middle range of other expert calculations.

The information Daponte gave to Robert Burns, an Associated Press reporter who called her in January, would have been available to anyone who came to her office and asked for the Iraqi folder for the "World Population 1992" handbook. Daponte said the file disappeared from her desk shortly after Burns's story appeared in The Washington Post and is still missing.

Hobbs and another supervisor later rewrote and released Daponte's report, reducing the number of direct, wartime civilian deaths from 13,000 to 5,000 and eliminating a Daponte chart breaking down the figures for men, women and children.

"I think it's rather scary that if an employee releases public information to the public, they can get fired for it," Daponte said. "My salary had been paid by tax dollars. I thought the public was entitled to know what we had come up with."