Voter registration, it is widely argued, raises the costs of voting, thereby decreasing turnout. Studies of turnout across states find that states with later registration dates or election day registration have much higher turnout rates. Eliminating registration barriers altogether is estimated to raise voter participation rates by 5 to 10 percentage points. This paper presents panel estimates of the effects of the introduction of registration that exploits changes in registration laws and turnout within counties. New York imposed registration on all of its counties in 1965; Ohio imposed registration in all of its counties in 1977. We estimate that the imposition of registration on counties that did not have registration in these states decreased participation over the long-term by 3 to 4 percentage points. Though significant, this is lower than estimates of the effects of registration from cross-sectional studies.