Software maintenance and evolution (SME) is an important but challenging topic area for university-level computer science education. Seminars can be used to provide students with versatile and up-to-date knowledge on scientifically relevant issues. We organized three systematic university-level seminars on SME. In these seminars 127 groups have each been assigned the task of analyzing one scientific SME article. The main results concern background factors relating to the students, groups and articles as these affect student success in the seminars. This paper presents a strict statistical analysis and a discussion of these factors. Fourteen hypotheses were set and tested regarding the relation of various background factors and a student's success in the seminar. The results indicate a clear relation between some of the factors and success. Most of the student-and group-related factors clearly affected student success, whereas most of the articlerelated factors did not. The study also revealed many important ancillary results. The results support organizing, studying, and improving feasible seminars in this area.