In order to give every student a fair chance to complete an academic study at a proper level and in a proper time institutes should be able to predict correctly how students will achieve in distinctive study phases. In particular it is important to identify early and reliably which students will fail to complete their study and why. Such an early identification may create the possibility to refer those students to another study or to offer them corrective measures to solve their mismatch with the learning environment. The goal of the studies in this thesis is to find factors to predict more reliably which subgroups of students will encounter what kind of problems in which study phase. To that purpose, the relationship between factors related to the student, the curriculum and the social environment on one side and student achievement on the other is investigated for both distinctive study phases and subgroups of students. One of the main outcomes of this thesis is a very useful model for the early and reliable prediction of students who will fail to pass the first-year curriculum within two years of study. By using this model we are able to identify at 6 months from the start a manageable group of students for a short remedial support programme. It is also shown that too many of our students would be dismissed on non-rational grounds at the end of the first year of study, if we would only use their number of credits obtained as selection criterion. Based on this latter finding the Binding Study Advice regulation in our medical school has been adapted. Gerard J.A. Baars studied Applied Educational Sciences at the University of Twente. Since August 2000, he works as a senior educational consultant at Risbo, Erasmus University Rotterdam.