The study addresses the following two questions: What is the contribution of modelling to fragmentation research? Which steps between problem recognition and problem solution are facilitated by modelling and which are not? Potential and limitations of models are demonstrated by two examples from the field of landscape fragmentation. Both examples address the problem of habitat fragmentation due to roads by analysing their effect on territorial wildlife populations. This paper shows that models can neither set normative goals nor values, but can contribute to a scientific justification of conservation goals and can support decision-making. The two examples give rise to the conclusion that fragmentation research has to be embedded in a (landscape) ecological context.