The college-educated are more likely to vote than are those with less education. Prior research suggests that the effect of college attendance on voting operates directly, by increasing an individual’s interest and engagement in politics through social networks or human capital accumulation. College may also increase voting indirectly by leading to degree attainment and increasing socioeconomic status, thus facilitating political participation. However, few studies have empirically tested these direct and indirect pathways or examined how these effects vary across individuals. To bridge this gap, we employ a nonparametric causal mediation analysis to examine the total, direct, and indirect effects of college attendance on voting and how these effects differ across individuals with different propensities of attending college. Using data from the 1979 and 1997 cohorts of National Longitudinal Surveys of Youth, we find large direct effects of college on self-reported voting and comparably smaller indirect effects that operate through degree completion and socioeconomic attainment. We find the largest impact of college on voting for individuals unlikely to attend, a pattern due primarily to heterogeneity in the direct effect of college. Our findings suggest that civic returns to college are not contingent upon degree completion or socioeconomic returns. An exclusive focus on the economic returns to college can mask the broader societal benefits of expanding higher education to disadvantaged youth.