The number of states with election-day registration (EDR) of voters doubled
in the early 1990s, providing a new opportunity to estimate the turnout im
pact of EDR. Because of some important and neglected features of the first
wave of EDR states, who adopted EDR in the early 1970s, there is good reaso
n to expect this second wave to generate larger estimates of EDR's turnout
impact. Controlling for other factors, new EDR programs are associated with
a turnout increase of about 6 percentage points in the midterm elections (
1990-1994), and 3 percentage points in the presidential elections (1992-199
6). Contrary to expectations, these estimates from the second wave of EDR s
tates do not exceed those generated by studies of the first wave of EDR ado
ption.