The use of hypermedia environments is increasing in school education. The interactivity in hypermedia environments challenges learners to autonomously navigate such environments. Particularly in multiperspective hypermedia environments (MHEs), which emphasize the multiperspectivity of a topic, it is important to apply navigational behaviors that take advantage of this information structure through which different perspectives can be selected and compared. However, we argue that and test whether the availability of sufficient spatial working memory (WM) resources is an important precondition for effectively engaging in this type of perspective processing. Specifically, we examined N = 97 German fourth-graders' navigational behaviors (i.e., perspective processing, content processing, and irrelevant processing) during hypermedia learning and their relation to spatial WM and performance. To this end, we developed an MHE on the biodiversity of 24 fish species. Participants’ navigational behavior was determined via log file analyses, while their performance was assessed by exploration questions and inferential questions (about fish) and by scientific transfer questions (about the biodiversity of another species). Our results indicated that spatial WM was positively related to perspective processing, which was positively related to performance. Mediation analyses revealed that perspective processing partially explained the relation between spatial WM and performance. Thus, both spatial WM and perspective processing are important for benefitting from MHEs.