The study of compositional effects is a central interest of sociologists in the field of educational research. In particular, the positive effect of the average achievement level of classes and ability groups on individual achievements of students has attracted much attention. This article presents a rational-choice theoretical model that (1) explains this compositional effect, (2) explains the differential effect of ability grouping on the achievements of low-and high-achieving students, and (3) predicts that the compositional effect is larger for low-achieving than for high-achieving students. By introducing new assumptions the model can be used to produce new hypotheses. This is illustrated by a derivation of hypotheses about the effects of different disciplinary climates on schools. It is concluded that a rational choice model can provide explanatory mechanisms currently lacking in research on compositional effects.