Problem solving in physics requires a certain quantity of knowledge of the subject matter: principles, procedures, etc. In addition, the problem solver must be able to access these principles and procedures in a given situation. Investigations have shown that failure in problem solving is often caused by lack of availability of knowledge, and also that availability is closely related to the organization of knowledge in memory. Opinions differ, however, on whether the optimal form of this organization should be centered around problem types or arranged in a hierarchical way. In this study two concrete examples of knowledge structures in the field of electricity and magnetism are compared. An experiment is also described, in which the actual knowledge structure of beginning students was studied. The outcome indicates that students with good results in problem solving organize their knowledge more in accordance with problem types than do students with poor results. The results of the experiment are discussed in the light of the two knowledge structures described. The possible role of these structures in physics teaching is treated in the final paragraph.