Assigning students to different classes on the basis of their achievement levels (tracking, streaming, or ability grouping) is an extensively used strategy with widely debated consequences. The authors developed a model of the effects of tracking on self-concept and interest that integrates the opposing predictions of "assimilation" and "contrast" effects, which specifies teacher-assigned grades as a major mediating variable, and tested it in 2 settings in which track level is clearly associated with different status-systematic tracking as a function of school type (Study 1, N = 14,341 German 9th-grade students) and separate streams within a comprehensive school system (Study 2, N = 3,243 German 9th-grade students). The results support predictions that students' math self-concept and math interest differ as a function of the achievement of their reference group, their own achievement, and their teacher-assigned grades. No systematic association between track level and math self-concept was found once individual student achievement, school-/stream-average achievement, and teacher-assigned grades were controlled.