When the long lines began to snake out of the polling station in the Montgomery County senior community Leisure World, County Council member Steven A. Silverman wondered whether elderly voters were having trouble with the state's new computerized voting machines.

"There were many people there who told me they were not computer literate, that they don't even use ATMs," said Silverman (D-At Large).

Silverman's hunch now appears to be supported by a comprehensive study of voters' first brush with computerized voting machines, used for the first time in four Maryland counties: Montgomery, Prince George's, Dorchester and Allegany.

Exit polling by University of Maryland researchers found that one-third of voters who lacked a college education needed help to use the machines; one-quarter of those who said they had infrequently used computers sought help with the machines; and about one-sixth of all voters said they got assistance from election judges.

"What this signals is that there are real privacy issues that need to be dealt with when voters are using the computerized machines, and there are serious usability issues that are extremely important," said Paul Herrnson, who heads the University of Maryland's Center for American Politics and Citizenship and was co-director of the study.

The General Assembly decided last year to overhaul Maryland's voting system in response to the Florida ballot count that delayed the results of the 2000 presidential election for weeks.

State budget officials decided to replace outdated systems in Montgomery, Prince George's, Allegany and Dorchester counties for this year's elections and to install computerized voting machines statewide by 2006.

The four counties and the state split the $15 million cost of the new AccuVote-TS system, manufactured by Global/Diebold Election Systems in Ohio.

Although the machines' inaugural run was relatively smooth in the other counties, confusion reigned in Montgomery during the September primaries. Several polling places opened late because the equipment was not set up. Inaccurate results were posted on the county Web site, while judges struggled through complicated forms and tabulations.

Instead of sending results by modem, poll workers had to drive computer disks across the county to the elections board in Rockville and faced long lines upon arrival. And nearly 100 machines were delivered whole to the county board by frustrated poll workers.

The University of Maryland study, however, focused on the ease of voter use and tried to gauge how voters responded to the new machines.

In addition to the voters who needed help to complete filling out their ballots, the study found that a small percentage of the machines -- about 3 percent -- malfunctioned in some manner.

Most of those problems, the study reported, involved card readers used to activate the machines. Election judges in some precincts addressed the problem by inserting the card for voters. Some voters also reported trouble navigating from screen to screen, especially in the "ballot review" section.

Herrnson said that while 3 percent sounds like a small figure, Florida's 2000 presidential election debacle was a reminder that perfection should be the goal.

"In several House races in Maryland, there was voter turnout of 200,000, so in those cases, 3 percent starts to sound like a pretty big number," he said.

Herrnson said he witnessed some of the trouble voters were having when he went to the polls in his hometown of Hyattsville.

"After I voted, I sat down and watched as people raised their hand while they were voting to ask for help," he said.

He said the study recently was submitted to the state Board of Elections for review, and he hopes that as the machines become more common, the board will devote more time to educating the public about how to use them.

Margaret Jurgensen, election director for Montgomery County, said she found the machines to be popular with voters and was pleased with the initial experience. She said her staff will continue to emphasize training of poll workers as well as programs to educate voters on the new machines. "This is the first year this has been deployed," she said. "There's always going to be a learning curve."

State board officials could not be reached to comment on the study, but on Election Day last month they defended the performance of the new machines, saying the few glitches should be expected with a new product.

And in Montgomery County, while some voters had trouble, the problems were far from universal.

"It was wonderful, and I'm not even computer literate," Winnie Soltz, 62, told a reporter. The Gaithersburg substitute teacher had just cast her ballot, without any difficulty at all.