Effects of full-time ability grouping on students' academic self-concept (ASC) and mathematics achievement were investigated in the first 3 years of secondary school (four waves of measurement; students' average age at first wave: 10.5 years). Students were primarily from middle and upper class families living in southern Germany. The study sample comprised 148 (60% male) students from 14 gifted classes and 148 (57% male) students from 25 regular classes (matched by propensity score matching). Data analyses involved multilevel and latent growth curve analyses. Findings revealed no evidence for contrast effects of class-average achievement or assimilation effects of class type on students' ASC. ASC remained stable over time. Students in gifted classes showed higher achievement gains than students in regular classes.