Earlier research has found that it is useful to distinguish situations in which students construct external representations on their own from situations in which they are expected to interpret already provided external representations. One type of representations that is particularly important for teaching mechanics is the free-body diagram. In this study, we investigated how inclusion of free-body diagrams into problem statements influences students' performance in solving mechanics problems. To that end two versions of a five-problem assessment instrument that only differed with respect to the inclusion/non-inclusion of free-body diagrams (FBDs) were administered to two groups of first year physics students. It was found that inclusion of free-body diagrams into the problem statements not only did not facilitate problem solving, but also impeded it significantly. Particularly large between group differences, in favor of the group not provided with FBDs, were detected for problems that required use of free-body diagrams showing resolution of forces into components. The results of our study indicate that consistency between internal and external representations of knowledge is a very important requirement for effective problem solving and effective learning of physics, in general. This consistency is most easily established when students use self-constructed external representations.