The November 2004 election was the first time all states were required to give provisional ballots to voters whose names did not appear on the voter registration list on Election Day. The results were mixed: although 1.6 million provisional ballots were cast and nearly 1.1 million were counted, standards differed considerably both between and within states, and many poll workers were apparently unfamiliar with the new rules. The Election Assistance Commission and electionline.org frame the issue with results from their recently published post-election surveys.

A variety of factors, including record numbers of new registrants and the first presidential election test of several voter ID laws, may produce a large number of provisional ballots in November. A new report from the Advancement Project examines the rejection rates of these ballots in 2006 and makes recommendations for reform, while a paper from David Andersen at Rutgers investigates the effect of ID requirements on the number of provisionals cast.

We launched the AEI-Brookings Election Reform Project in June 2005 with the encouragement and financial support of the John S. and James L. Knight Foundation. Five years later we bring the project to a close. We take this opportunity to reflect on the state of election administration in the United States almost a decade after the extended and controversial Florida vote count in the 2000 presidential election and suggest how additional changes in technology, election law and administrative practices might further strengthen American elections in the years ahead.