In both 2008 and 2012, about one-third of U.S. voters cast their ballots before Election Day. Reformers have argued that early voting lowers participation costs and should therefore increase turnout. Recent research, however, has reported that no positive relationship exists. The literature widely omits consideration of important differences in early voting implementation within states. I break from past research and measure early voting availability at the county level, where it often varies considerably. I rely on Election Assistance Commission data on the number of early voting sites available in 2008 and 2012. Specifically, I measure the effect of a county's early voting site density on turnout. My model controls for other known participation predictors, including lagged turnout, demographics, political variables, and voter identification requirements. Ultimately, I find that early voting site density has a significant and positive effect on voter turnout.