Existing research finds minimal gains from attending elite US secondary schools. This paper estimates the causal effect of attending a selective public boarding school, an institutional model increasingly used by states to serve academically gifted students. Regression discontinuity estimates using multiple admissions thresholds show math score gains and college application and enrollment patterns that shift away from less competitive colleges. Effects are concentrated among minorities, students with lower prior individual achievement, from rural neighborhoods, or lower-achieving sending schools. The opportunity to attend selective boarding schools reduces the tendency of disadvantaged or under-represented students to attend a less selective college by at least one-quarter. [This is the in press version of an article published in "Economics of Education Review."]