Few doubt that traffic congestion has enormous financial and environmental costs, and that transit systems like MARTA ease some of that. The question is how much.

Mass transit advocates demonstrated at the capitol Monday and released a report estimating MARTA saves metro Atlantans $220 million annually in gas costs and keeps 643,000 metric tons of carbon dioxide out of the air.

National travel analysts called a core assumption of the study into question.

According to the consumer and environmental group PIRG, the savings from MARTA come from reducing driving and congestion. To come to those conclusions it assumes that if MARTA disappeared, every passenger would take the same trips by vehicle. They would sometimes share rides, so the report assumes an average of about 1.5 riders per vehicle.

The trouble is, as far as the study authors know, no one's ever shown how people who lose transit change their travel habits: how many people would give up some trips, walk, or have friends chauffer them, or buy a car and drive more miles than they rode on transit. The study says it chose the mile-for-mile replacement "for the sake of simplicity."

"At the national level we know that in 2001 about 46 percent of transit trips were made by folks without a household car," Steven E. Polzin, director of the Transit Research Program at the University of South Florida, wrote in an e-mail. "Since we won't know what would actually happen in the short term or long term if transit went away or were cut way back it gives folks some liberty on characterizing the impacts or benefits of transit."

Alan Pisarski, author of the nation's premier commuting study, Commuting in America, called the PIRG report's assumption "self-serving."

To make its assumption, PIRG cites another report, also done for transit advocates, by the research firm ICF. The author of the ICF report, Linda Bailey, said many complicated factors would affect people's transportation choices if transit disappeared, and "I'm not sure how you would study it." In the absence of such data, she said, she believed her assumptions were conservative.

Tony Dutzik, a policy analyst for the Frontier Group and a co-author of the PIRG report, said, "I would welcome somebody to come in so we could answer that question a little bit better."

He said that his group's assumption was conservative compared to others. They used a gas cost of $2.68 a gallon to make their estimated.

"I'm confident we've got it, if not exactly right, then close to right," Dutzik said.