Do schools reduce or perpetuate inequality by race and family income? Most studies conclude that schools play only a small role in explaining socioeconomic and racial disparities in educational outcomes, but they usually draw this conclusion based solely on test scores. We reconsider this finding using longitudinal data on test scores and four-year college attendance among high school students in Massachusetts and Texas. We show that unexplained differences between high schools are larger for college attendance than for test scores. These differences are arguably caused by differences between the schools themselves. Furthermore, while these apparent differences in high school effectiveness increase income disparities in college attendance, they reduce racial disparities. Social scientists concerned with schools’ role in transmitting inequality across generations should reconsider the assumption that schools either increase or reduce all disparities and should direct attention to explaining why high schools’ effects on specific outcomes and groups of students appear to vary so much.