Estimating the impact of initial college choice on student outcomes is confounded by the non-random nature of college selection. We identify two new contexts where admission test score thresholds provide exogenous variation in access to four-year public colleges. We study Georgia’s state university system, whose thresholds are known to students, as well as other colleges whose threshold use is hidden from students but can be detected in our data. Using the universe of SAT takers in the high school classes of 2004-07, a regression discontinuity design comparing the relatively low-skilled students just above and below these thresholds yields two main findings. First, access to these four-year public colleges diverts students from two-year colleges or lower quality four-year colleges. Second, access substantially increases bachelor’s degree completion rates, particularly for low-income students. This is strong evidence of a college choice effect, as access allows students to attend four-year colleges from which some are capable of graduating.