A growing number of people rely solely on cell phones to make and take calls, putting them out of reach of polling organizations trying to get a fix on the American electorate.

Many cell-only users are young and mobile, a demographic that often doesn't vote. That makes survey researchers confident their polling, which excludes cell phone numbers, reflects the opinions of those likely to have an impact on Election Day. Still, with reports of unprecedented voter registration, many young voters could be flying under the pollsters' radar.

"Pollsters don't think the cell phone issue will affect them this year, but they are worried about it," said Michael Brick, a survey methods specialist at Westat, a research firm in Rockville, Md. "This may be the last round of presidential elections before it does have an effect."

When tracking this year's election, pollsters contact people on traditional phones. About 5 percent of all households receive telephone service only by cellular phone, according to a face-to-face survey done earlier this year by the Census Bureau and the Bureau of Labor Statistics. Among young adults up to age 24, the number is close to three times as high.

"Many of these people are not voters," said Linda Piekarski, vice president of database and research at Survey Sampling International, which provides samples for the research industry. "They've always been hard to get into our polls anyway. They tend to be non-responsive."

Survey researchers have much more to learn about this group, however, especially its demographic and political inclinations. An AP-Ipsos poll taken last week showed President Bush and Sen. John Kerry were running about even among voters 18-29, though Kerry has led among that group in some polls.

Only about three in 10 adults age 18-24 voted in the last two presidential elections, said Curtis Gans, director of the Committee for the Study of the American Electorate. That number spiked to almost four in 10 in 1992.

Survey researchers are prohibited from using automated dialing equipment to call wireless numbers and can be fined for each infraction. Known wireless prefixes and blocks of numbers used for wireless phones are removed from samples used in random-digit dialing, the dominant method of contacting respondents for polls.

The American Association for Public Opinion Research is studying ways to include this group in surveys in the future because it is likely to grow larger.

"It's a manageable problem at this point because the size of the population is so small," said Andrew Kohut, director of the Pew Research Center for the People & the Press. "In a very close election it could be a factor, but they would have to lean overwhelmingly one way or the other."

Partisan pollsters say other problems in campaign polling are bigger concerns. GOP pollster Bill McInturff said pollsters might eventually have to move to other techniques that use combinations of random-digit dialing and the Internet.

"We may get bitten in the rear this election," said Democratic pollster Celinda Lake. "But the bigger problems are voter turnout models used and machines that keep people from picking up their phones."

Most in the business agree the cell-phone-only phenomenon will eventually affect telephone polling.

"I think it's unlikely we're going to see a 'Dewey defeats Truman' event this year," said Philip Trounstine, director of the Survey and Policy Research Institute at San Jose State University. In 1948, the Chicago Daily Tribune erroneously declared Republican Thomas Dewey the winner over President Harry Truman in part because polls had long favored the challenger.

Yet, Trounstine said, "People in the polling business can't take an ostrich attitude about this."