This dissertation examines the influence of blended learning on student's learning behavior with respect to student heterogeneity. The analysis is based on theoretical models. Since blended learning offers a variety of learning opportunities which are available around-the-clock, it is able to influence learning in a different way than traditional teaching methods. Especially online feedback helps students to self-regulate learning and to assess their abilities. Taking the growing heterogeneity of students into consideration, blended learning may be a measure to rethink homogeneous university structures. Chapter 1 and Chapter 2 follow Göcke (2002) by applying his model of learning by doing to the framework of university courses. These chapters analyze how learning behavior is influenced by online learning opportunities. Chapter 3 introduces a broader perspective, shedding light on the impact of blended learning on student heterogeneity which characterizes today's higher education system. Dynamic utility maximization in Chapter 1 emphasizes that around-the-clock online learning opportunities lead to a nearly constant learning activity in time. In this connection, we also show that constant learning increases the learning progress. Furthermore, the introduction of an online mid-term test leads to a higher level of learning time before the test and consequently shifts the learning activity towards the beginning of the semester. Since studies on flawed self-perception indicate that students are not able to self-assess themselves accurately, Chapter 2 assumes imperfect knowledge about the individual ability. As a result, we are able to show that online exercises and a mid-term test improve self-assessment of skills and the failure rate of the exam if students persistently make use of these feedback sources. Chapter 3 introduces a new perspective by looking at blended learning and the higher education system. On the one hand, Chapter 3 shows that blended learning guides the self-study of students and improves flawed self-assessment and on the other hand, blended learning is able to transform teaching routines in higher education to adapt to the heterogeneous student body.