Krylov subspace methods have become very popular, not only for solving large scale linear systems, but also in the area of model order reduction. It is also well known that the performance of iterative solution techniques depends crucially on the choice of preconditioning matrix. Within the area of model order reduction, the concept of preconditioning has not yet been introduced. However, recent reduction problems arising in electronics applications clearly demonstrate the need for some kind of preconditioning step prior to the actual model order reduction. In this paper, we present initial ideas and some test results.