In a few months, Americans will cast ballots in the earliest and most crowded presidential primary season in our history. Many will arrive at the polls only to find they are not on the list of registered voters.

In this, and in every election, the accuracy of voter registration lists is essential.

In states throughout the country, the integrity of voter registration lists is in question. In 2004, a study by the Chicago Tribune found that more than 181,000 of the voters on the rolls in six swing states were deceased.

Registration rolls are created piecemeal, relying on local registrars, state motor vehicle agencies and a wide array of nonpartisan and partisan get-outthe- vote campaigns. Efforts to ensure that registration rolls are up to date are impeded by limitations in state data management technologies and federal legal constraints on when names can be removed from lists.

And while the private sector has developed efficient ways to keep consumer records current, the 40 million Americans who move each year must manually reregister to vote when they change addresses.

Democracies around the world are doing a far better job than we are at ensuring the integrity of their voter rolls. Perhaps the most important step they have taken is in the use of automatic registration practices to compile and maintain their lists.

In Canada, a linked system of government agency databases registers eligible voters as soon as they reach voting age. Here at home, states are experimenting with similar innovations:

• Arizona was the first state to allow online registration, and in 2006 more than half of all new registrants did so via the Internet.

• Florida just updated its laws to allow new drivers to preregister when they get their first license. When these new drivers turn 18 they will automatically be added to the voter rolls.

• Ohio is undertaking an effort to register those who relocate frequently by tying registration to change-of-address systems.

Each of these innovations represents a promising step forward, and in the coming election cycle they deserve serious evaluation. But even if they succeed, they are only Band-Aids.

It is time to take an entirely new look at the way we compile and maintain our voter registration rolls.

It is time to take a hard look at whether universal portable registration is possible in the United States — a system where states would have a comprehensive list of all their voters, registration would seamlessly follow those who move, ineligible names could not be added to the list and information would be managed reliably.

With solid evidence collected during the 2008 election cycle, it will be time to wrestle with that question and other bold ideas that would transform rather than repair our elections.

Michael Caudell-Feagan is project director of Make Voting Work, an initiative of the Pew Center on the States.