The number of Americans infected by the virus that causes AIDS each year is much higher than the government previously estimated, U.S. health officials said Saturday, acknowledging that they had understated the level of the epidemic.

The United States is now thought to have had roughly 56,300 new HIV infections in 2006, a dramatic increase from the 40,000-a-year estimate used for the last dozen years. The new figure is because of a better blood test and new statistical methods, and not a worsening of the epidemic, officials said.

It was not immediately clear what the American change implies for the accuracy of Canadian statistics. Canadian officials have counted about 2,500 new reported HIV infections annually in recent years.

The revised figure from the U.S. Centers for Disease Control and Prevention and the methodology behind it were to be presented Sunday, the opening day of an international AIDS conference in Mexico City.

Since AIDS surfaced in 1981, health officials have struggled to estimate how many people are infected each year. It can take a decade or more for an infection to cause symptoms and illness.

One expert likened the new estimate to adding a good speedometer to a car. Scientists had a good general idea of where the epidemic was going; this provides a better understanding of how fast it's moving right now, said the expert, David Holtgrave of Johns Hopkins University.

Based on the new calculations, officials believe annual HIV infections have been hovering around 55,000 for several years in the United States.

"This is the most reliable estimate we've had since the beginning of the epidemic," said Dr. Julie Gerberding, the CDC's director. She said other countries may adopt the agency's methodology.

According to current estimates, around 1.1 million Americans are living with the AIDS virus. Officials plan to update that number with the new calculations, but don't think it will change dramatically, a CDC spokeswoman said.

The new infection estimate is based on a blood test that for the first time can tell how recently an HIV infection occurred. Past tests could only detect the presence of HIV, so determining which year an infection took place was guesswork — guesswork upon which the old 40,000 estimate was based.