The new testing method, according to Stephen L. Johnson, the E.P.A. administrator, would come much closer to bridging "the gap between what the window sticker says and what consumers can expect in their fuel economy."
The agency expects to introduce the changes starting with 2008 models, which will go on sale as early as a year from now.
Consumer groups have long complained that the E.P.A.'s ratings are far too optimistic, compared with fuel economy that drivers achieve under real-world conditions. Recent tests by Consumer Reports magazine, for example, found that E.P.A. window stickers could be off by up to 50 percent.
The new calculations will have the greatest impact on hybrid-electric vehicles, the agency said, cutting estimated fuel economy sharply on some of the industry's most sought-after models now that gasoline prices have soared.
For all vehicles, the agency said its new testing methods would result in a 10 to 20 percent drop in fuel-economy estimates in city driving, and a 5 to 15 percent decline in highway performance.
But for hybrids, which run off a gasoline engine and an electric battery, city driving estimates could drop 20 to 30 percent. The decline in highway ratings would be 5 to 15 percent, the same as for regular cars.
Buyers are willingly paying thousands of dollars above the price of conventional vehicles, and waiting up to a year in the case of Toyota's most popular hybrid, the Prius, all in the belief that they yield much better gas mileage - the Prius officially gets 60 m.p.g. in the city and 51 on the highway.
While hybrids are almost always more fuel efficient than conventional vehicles, E.P.A. officials said their estimates for city driving would shrink more because their engines were more sensitive to changes in road conditions, as well as the use of fuel-draining features like air-conditioning and electronic controls.
The E.P.A. could not give estimates for specific vehicles. And at this point, they would not be completely accurate. Automakers have some time to improve their fuel economy ratings under the new calculations.
Mr. Johnson of the E.P.A. said the proposed standards, which will be open to public comment for 60 days, are meant to depict more accurately what consumers can expect from new cars and trucks. In its report in October, Consumer Reports said it found shortfalls in as many as 90 percent of the 203 vehicles it tested, which were built in the model years from 2000 to 2006.
The most inaccurate results came in city driving, especially involving hybrid cars. But popular models like the Chrysler 300C sedan also had inaccurate estimates: the magazine calculated its fuel economy in city driving at 10 miles a gallon, versus the 17 m.p.g. estimate on its window sticker, a difference of 41 percent.
Likewise, the E.P.A. measured the city driving performance of the Honda Odyssey minivan at 20 m.p.g., but the magazine said it achieved only 12 m.p.g., a 40 percent discrepancy.
The agency's new testing method, however, will have no effect on the regulations used in judging whether auto companies are meeting their fuel economy standards, which are overseen by the Transportation Department. The new government estimates, like those used by the consumer magazine in its testing, will include much more information to make the calculations.
For example, the E.P.A. now will use data from tests meant to show performance under conditions known to deplete gasoline. These include driving at high speeds and with rapid acceleration, as well as when air-conditioning is in use and when the weather is cold.
The agency also said it planned to consider other conditions that could hurt fuel economy, including road grade, wind velocity, whether tires were properly inflated and the type of fuel in a vehicle's gas tank.
Mr. Johnson stressed that fuel economy estimates did not reflect anything that auto companies - or consumers - have done wrong when it comes to gasoline mileage.
"What this proposal does is give consumers better information," Mr. Johnson said. He added that "there is no perfect test" and that consumers' performance would vary, even once the new ratings are in effect.
Also set to change, assuming the proposal is adopted, is the appearance of window stickers themselves. The agency is considering four designs for the stickers.
Detroit's two biggest auto companies, General Motors and Ford Motor, said Tuesday that they were in favor of more accurate information, but did not say whether they would embrace the new calculations.
"G.M. supports providing consumers with more accurate fuel economy data for comparative purposes, and we appreciate the opportunity to provide comments on this rulemaking," the company said in a statement.
Ford said it backed changes "that will help provide consumers with more meaningful information for their purchasing decisions."
Martha Voss, a Toyota spokeswoman based in Washington, said: "Circumstances have changed in the world since the rules were established. We want customers to get the most accurate information possible. The ratings might change, but the performance of our customers' car won't change."
Environmental groups said the move was a much-needed first step, but it did little to deal with the fundamental issue of improving fuel economy itself.
"The current fuel economy labeling system is broken," said Don MacKenzie, an engineer with the Union of Concerned Scientists.
"E.P.A.'s proposal is a long overdue tuneup that better reflects the growing diversity of vehicle technologies and today's driving conditions," he added.
Dan Becker, director of the Sierra Club's global warming program, said the new government standards would not address how to increase the number of fuel-efficient vehicles.
"The real issue," Mr. Becker said, "is making auto companies put the technology in cars that make them go farther on a gallon of gas."