First, the presentation will give a brief overview over the data of the initial survey 1969/70 and the three follow-ups 1985/86, 1996/97 and 2010 of the Cologne High School Panel (ChiSP) and has a differentiated look at causes for panel mortality at the three stages. In a series of regression models then we will examine how well different kinds of drop-outs can be explained by indicators of biographical success and the interest in the topics of the survey. The results justify the differentiated analysis and show that different causes for the loss of participants in the ChiSP require different explanations. A main finding is that in a life course study like ours the willingness to participate is influenced by biographical success, especially educational success. The selective loss of participants leads to a small success bias, which can be seen at indicators for educational success and its correlates. As the initial sample includes only high school students, it is very selective from the beginning. Its selectivity increases over the course of 40 years a little, but it will be shown that this does not affect the results of typical models of the sociology of education substantially.