In this work we discuss experiences we gained during several years of teaching students of engineering multibody system analysis [1], [2]. Our teaching is organized in two actions: classroom lectures and laboratory activities, where students must model, simulate, and analyze several multibody systems of growing complexity. In our paper we will report and comment the second action. Through our experience of evaluating the students’ laboratory work, we have recognized the importance of the errors that students do in modeling and simulating a system. Obviously, we do not take into consideration many trivial errors concerning the use of the software that we have adopted to perform the laboratory activities (DADS), but only conceptual errors that show how a model that “seems to work” can be a poor model. Many of these errors can help teachers to understand what is wrong in students comprehension both of the problems they must and the simulation features, and to orient the teaching plan