The German school system is traditionally characterized by a variety of homogenization mechanisms aimed at reducing performance differences. On the other hand, performance heterogeneity is seen as an opportunity for teaching as well as for the performance-related and social development of children. This work addresses the question of whether school learning can be more successful in performance-homogeneous or heterogeneous classes and whether the type of class composition is equally significant for all students. The data basis is provided by the Hamburg longitudinal study "Competencies and Attitudes of Students (KESS)", which depicts the learning development of an entire student cohort over three measurement points from grade 4 to 8. Multilevel analyses show that a performance-heterogeneous class composition is not fundamentally associated with disadvantages.