New government estimates of the number of Americans who become infected with the AIDS virus each year are 50 percent higher than previous calculations suggested, sources said yesterday.

For more than a decade, epidemiologists at the Centers for Disease Control and Prevention have pegged the number of new HIV infections each year at 40,000. They now believe it is between 55,000 and 60,000.

The higher estimate is the product of a new method of testing blood samples that can identify those who were infected within the previous five months. With a way to distinguish recent infections from long-standing ones, epidemiologists can then estimate how many new infections are appearing nationwide each month or year.

The higher estimate is based on data from 19 states and large cities that have been extrapolated to the nation as a whole.

The CDC has not announced the new estimate, but two people in direct contact with the scientists preparing it confirmed it yesterday.

What is uncertain is whether the American HIV epidemic is growing or is simply larger than anyone thought. It will take two more years of using the more accurate method of estimation to spot a trend and answer that question.

"The likelihood is that this bigger number represents a clearer picture of what has been there for the past few years. But we won't know for sure for a while," said Walt Senterfitt, an epidemiologist who is the chairman of the Community HIV/AIDS Mobilization Project (CHAMP), a New York-based activist organization.

There is evidence, however, that at least some of the higher number may reflect an uptick in infections in recent years. Information from 33 states with the most precise form of reporting showed a 13 percent increase in HIV infections in homosexual men from 2001 to 2005.

Ironically, the news comes less than two weeks after UNAIDS, the United Nations agency responsible for charting the course of the global epidemic, drastically reduced its estimate of the number of people living with the disease worldwide from 40 million to 33 million. The reason was the same: Crude methods of counting were replaced by better ones.

"People in the United States are under the impression that this is more of an international than a domestic issue," said Rowena Johnston, vice president for research at amfAR, an AIDS research foundation. "Yet these new CDC numbers are telling us that not only does this continue to be a serious problem, it is actually a larger one than we suspected."

A study describing the new U.S. estimate is under review at a scientific journal, Thomas W. Skinner, a CDC spokesman, said last night.

"We have to wait until this paper comes out, until it has gone through peer review, before we know what the new estimates look like," he said.

Rumors have circulated for weeks in newsletters and blogs that CDC, the federal government's principal epidemiology agency, was preparing a dramatic upward revision of HIV incidence. The Washington Blade, a gay-oriented newspaper, reported rumors of the new estimates two weeks ago.

The CDC has reported the figure of 40,000 new infections each year for more than a decade, citing it as evidence that the epidemic in this country is stable. But while widely quoted, that number has never been adequately explained or justified, in the eyes of many epidemiologists.

"There was skepticism about the validity of how that estimate was reached," said Rochelle Walensky, an infectious diseases physician and mathematical modeler at Harvard Medical School.

Some activists also were skeptical about it.

"It just doesn't seem plausible to me that it would be the same year after year," said Mark Harrington, executive director of Treatment Action Group, an AIDS activist think tank in New York.

Few doubt, however, that accurately counting new HIV infections is unusually difficult. About one-quarter of people infected with the virus do not know they are. The infection is largely "silent" for a decade in most people, and a substantial number go for testing only as they develop the symptoms of AIDS, the late stage of the illness.

Only recently has CDC put intense pressure on state and city health departments to report by name everyone who tests positive for HIV. Previously, health departments had to report only the people who had progressed to AIDS.

Counting only AIDS cases was an acceptable substitute for counting new infections in the era when AIDS treatment did not significantly prolong life. But with the arrival of combinations of potent antiretroviral drugs in 1995, AIDS patients began living years longer, making the estimates increasingly less accurate.

The new system in which health departments record individuals who have just tested positive for the first time will eventually provide a much clearer picture of the epidemic. However, some people oppose it, arguing that it will keep the potentially infected from coming in to be tested.

"There are so many barriers to testing and reporting," Harrington said. "We are grasping in the dark, as far as I am concerned, about the real size and shape of the epidemic."

The 19 states and cities that contributed the data for the new estimate include New York City, New York, New Jersey, Connecticut, Texas, Florida, and several Southern and Midwestern states.

The new method of estimating HIV incidence makes use of the observation that a person who is recently infected with HIV and whose immune system has just begun to make antibodies against the virus shows a weaker reaction in the standard AIDS blood test than those whose immune systems have been making antibodies for years.

By altering the test-tube conditions, scientists can identify those who react weakly -- and with them, the percentage of a batch of HIV tests that come from people newly infected.

The method is called the STAHRS method, for serological testing algorithm for recent HIV seroconversion.